JP2013167865A - Confidential information hiding device, confidential information restoring device, confidential information hiding program, and confidential information restoring program - Google Patents

Abstract

Using steganography based on morphing, data that cannot be morphed is concealed as cover data in cover data, the cover data is concealed from cover data, or a plurality of cover data is simultaneously transferred from the cover data. Concealed in a small number of cover data
The present invention relates to a steganography technique for generating a stego object s by concealing secret data m in cover data c, reference data, feature vectors of each reference data, contribution in each reference data, and secret. By applying a morphing technique for generating morphing data based on the data, the calculation means 15 generates a stego object that hides the secret data in the reference data, or restores the secret data from the stego object.
[Selection] Figure 2

Description

  The present invention relates to a morphing-based steganography technique, and more particularly, to a steganography technique for generating a stego object by concealing secret data in cover data, reference data, a feature vector of each reference data, and By applying a morphing technique for generating morphing data based on the contribution in each reference data and the secret data, a stego object concealing the secret data is generated or the secret data is restored from the stego object The present invention relates to a secret information concealment device, a secret information restoration device, a secret information concealment program, and a secret information restoration program.

Conventionally, by utilizing steganography (steganography technology), predetermined cover data c (cεC, c is an abbreviation of cover, and C is a set of all cover data) is transferred to secret data m ( There is known a technique of embedding mεM, m is an abbreviation of message, and M is a set of all secret data. Assuming that a function (hiding function) for embedding the secret data m in the cover data c is f, f becomes a function as shown in the following formula (1).

Here, s is a stego-object, sεC, and k _s is a stego-key. In formula (1), the element indicated by [] is optional and is not necessarily required. In order to increase safety, in conventional steganography, image data is used as the cover data c, and the stego object s is visually similar to the cover data c, and the cover data c once used is again used. It was necessary not to use it.

The secret data m embedded (hidden) in the stego object s can be reproduced by the extraction function f ⁻¹ using the cover data c, the stego object s, and the stego key k _s. . That is, the secret data m can be obtained by the following equation (2).
In order to increase security, the secret data m can be encrypted and then concealed (embedded) in the cover data c.

  On the other hand, a method of embedding secret data (covering method) in cover data by using morphing (morphing technology) has been proposed (see, for example, Patent Document 1 and Non-Patent Document 1).

As a method of embedding secret data, image morphing using image data, music morphing using music data, or the like can be used. In general, morphing can be expressed by the following relational expression (3).

In this relational expression (formula (3)), I ₁ and I ₂ are reference data, respectively, and F ₁ and F ₂ are reference data I ₁ and reference data I ₂ required for morphing. The feature vector is shown. Further, a and 1-a indicate the contributions of the reference data I ₁ and the reference data I ₂ , respectively, and I ₀ indicates data transformed (generated) as a result of morphing. Hereinafter referred to as morphing data _{I 0.}

In the case of image morphing, I ₀ , I ₁ and I ₂ are all two-dimensional images or three-dimensional images. In the case of music morphing, I ₀ , I ₁ and I ₂ are all numerical sequences. Here, it is assumed that I ₀ , I ₁ , and I ₂ are vectors having N _d elements without losing generality. Further, the feature vector is assumed to be a vector having N _f elements composed of feature point and feature line position (coordinate) information.

Using the morphing defined as described above, the relationship can be shown as follows when the morphing and the steganography are associated with each other.
Secret data m = reference data I ₁
Cover data c = reference data I ₂
Stego key k _s = {F ₁ , F ₂ , a}
Stego object s = morphing data I ₀

In this way, by associating morphing with steganography, it is possible to conceal reference data I ₁ as reference data I ₂ as secret data m (see, for example, Patent Document 2).

Of course, it is possible to conceal I ₂ from I ₁ by setting the secret data m as the reference data I ₂ and the cover data c as the reference data I ₁ . Hereinafter, in order to simplify the description, a case where I ₁ is concealed by I ₂ is shown as an example. The function f for embedding the secret data m = I ₁ in I ₂ can be realized as follows.

[Algorithm for realizing function f]
The output stego object s is designated as morphing data I _0, and the data required to obtain the morphing data I ₀ are secret data m = I ₁ , cover data c = I ₂ , stego key k _s = {F ₁ , F ₂ , a},
Input: m = I ₁ , c = I ₂ , k _s = {F ₁ , F ₂ , a}
Output: s = I ₀
Thus, the morphing data in which the secret data is concealed can be obtained by the following three steps.

STEP 1: The feature vector F ₀ of the morphing data I ₀ is obtained using the equation (4).
However, F _jp (j = 0, 1, 2) indicates the p-th element of F _j , and a indicates a real number satisfying 0 <a <1.

STEP 2: Transform secret data and reference data using equation (5).
Here, I ^w _j (j = 1, 2) is the transformed data of I _j , and W is the transformation (warping) function. I _j (j = 1, 2) is transformed by the transformation function W to generate I ^w _j (j = 1, 2). I ^w _j (j = 1, 2) has a feature vector F ₀ .

· STEP3: seek the morphing data _{I 0} in Equation (6).
However, I _0p , I ^w _1p , and I ^w _2p indicate p-th elements of I ₀ , I ^w ₁ , and I ^w ₂ , respectively.

By thus going through the procedure of STEP1～STEP3, reference data _{I 1} to the reference data _{I 2} it is possible to obtain the morphing data _{I 0} is a data hiding.

On the other hand, it is also possible to extract (restore and reproduce) the reference data I ₁ which is secret data from the morphing data I ₀ . The extraction function f ⁻¹ for extracting (restoring and reproducing) the secret data m = I ₁ from I ₀ can be realized as follows.

[Algorithm for Realizing Function f ⁻¹ ]
The secret data m to be output is referred to as reference data I _1, and the data required to obtain this reference data I ₁ are stego object s = morphing data I ₀ , cover data c = I ₂ , stego key k _s = If {F ₁ , F ₂ , a},
Input: s = I ₀ , c = I ₂ , k _s = {F ₁ , F ₂ , a}
Output: m = I ₁
Thus, the secret data m (reference data I ₁ ) can be extracted (restored and reproduced) from the morphing data I ₀ by the following four steps.

STEP 1: The feature vector F ₀ of the morphing data I ₀ is obtained using the equation (4).
However, F _jp (j = 0, 1, 2) indicates the p-th element of F _j , and a indicates a real number satisfying 0 <a <1.

STEP 2: Data I ^w ₂ obtained by transforming the cover data is obtained using equation (5 ′).
Here, I ^w ₂ indicates data obtained by modifying the reference data I ₂ , and W indicates a deformation (warping) function. I ₂ is deformed by the deformation function W to generate I ^w ₂ . I ^w ₂ has a feature vector F ₀ .

STEP3: The modified data I ^w ₁ of the secret data I ₁ is obtained using the equation (7).

STEP 4: Extract (restore, reproduce) secret data I ₁ using equation (8).

In this way, by applying the technique for generating morphing data to steganography, the morphing data I ₀ (stego object) is generated by concealing the reference data I ₁ (secret data) from the reference data I ₂ (cover data). can do. It is also possible to extract (restore and reproduce) the reference data I ₁ (secret data) from the morphing data I ₀ (stego object).

In steganography based on morphing, when the morphing data I ₀ (stego object s) is image data or music data, the image data or music data may be “natural”, and the cover data I _{2 is} not necessarily used. There is no need for visual or auditory matching. Therefore, by using steganography based on morphing, it is possible to conceal secret data (reference data I ₁ ) having a large data amount.

  Various methods are conceivable for realizing the deformation function W. For example, methods such as field warping and mesh warping can be used only for image morphing. Steganography based on morphing does not limit the method of realizing the deformation function W, and any method may be adopted as long as natural morphing data can be obtained.

JP 2007-219230 A JP 2009-171116 A

Satoshi Kondo, Takefuku Tsuji, Hideyuki Takagi, "Information hiding method using morphing technology", IEICE technical report, IEICE, February 13, 2006, Vol.105, No.608, pp .145-150

  By the way, as described above, in steganography based on morphing, it is necessary that the secret data be morphable data. For this reason, for example, in the case of simple numerical data that does not correspond to image data or music data, it is difficult to use steganography based on the morphing described above as it is.

Further, in the above-described steganography based on morphing, one reference data I ₁ (secret data) is hidden from one reference data I ₂ (cover data) by steganography based on morphing. For improvement, it is desired to conceal secret data using a plurality of reference data instead of using only one reference data as cover data.

  Further, the secret data is not necessarily limited to one type, and it is also required to conceal a plurality of types of secret elements in the same cover data.

  Furthermore, when the secret data is concealed in the image data based on the existing morphing, the image data (stego object) after concealment may be an unnatural image. The unnatural image data generally has a problem that it is easy for a third party to draw attention and the possibility of being attacked increases.

  The present invention has been made in view of such problems. As a first problem, secret data that cannot be morphed can be concealed in cover data by steganography based on morphing. As a second problem, In steganography based on morphing, secret data can be concealed using a plurality of cover data. As a third problem, in steganography based on morphing, a plurality of types of secret elements are concealed in the same cover data. As a fourth problem, a secret information concealment device, a secret information restoration device, and a secret information concealment program capable of concealing secret data so that a stego object does not become unnatural even if the secret data is concealed It is another object of the present invention to provide a secret information restoration program.

In order to solve the above problems, a secret information concealment device according to the present invention provides:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _n, different By applying a morphing technique for generating morphing data based on ν pieces (ν> n) of secret data m _{1 to} m _{ν obtained} by quantifying secret elements of the same type in the same format and a scaling factor A,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key The feature vector {F ₁ , F ₂ ,..., F _n }, the contribution {a ₁ , a ₂ ,..., a _n }, the scaling factor A, and the number of secret data ν are set as k _s. ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
And scale the secret data m _q (q = 1, 2,..., Ν),
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
and a stego object generating means for generating a stego object s in which v secret data m is concealed.

Moreover, the secret information concealment program of the secret information concealment apparatus according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _n, different By applying a morphing technique for generating morphing data based on ν pieces (ν> n) of secret data m _{1 to} m _{ν obtained} by quantifying secret elements of the same type in the same format and a scaling factor A,
A secret information concealment program of a secret information concealment device for realizing a method of generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key The feature vector {F ₁ , F ₂ ,..., F _n }, the contribution {a ₁ , a ₂ ,..., a _n }, the scaling factor A, and the number of secret data ν are set as k _s. ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
And scale the secret data m _q (q = 1, 2,..., Ν),
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
With seeking stego object s _p, the value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
It is a secret information concealment program of a secret information concealment device that realizes a stego object generation function for generating a stego object s in which ν secret data m is concealed.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The sum of the contributions satisfies a ₁ + a ₂ +... + A _n = 1, and the scaling factor A is a positive real number less than 1.

  By using the secret information concealment device and the secret information concealment program according to the present invention, if different types of secret elements are expressed by numerical values of the same format (for example, unsigned 8-bit integers), steganography based on morphing It is possible to hide the secret data in the reference data using.

Furthermore, the secret information restoration apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν and the scaling factor A,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, a scaling factor A, and the number of secret data ν, and the extracted secret data m is designated as secret data {m ₁ , m _{2 ,.} As
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
Is used to find the scaled secret data x _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. The process proceeds to the process of determining the above, and the determination process is repeatedly executed until q ≧ ν,
When q <ν is not satisfied,
By extracting ν secret data m _i (i = 1, 2,..., Ν) based on
And secret data reproduction means for extracting ν secret data m concealed in the stego object s.

Further, the secret information restoring program of the secret information restoring apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν and the scaling factor A,
A secret information restoration program of a secret information restoration device for realizing a method of restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, a scaling factor A, and the number of secret data ν, and the extracted secret data m is designated as secret data {m ₁ , m _{2 ,.} As
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To obtain the scaled secret data x _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. The process is shifted to the process of determining whether the determination process is repeated until q ≧ ν,
When q <ν is not satisfied,
By extracting ν secret data m _i (i = 1, 2,..., Ν) based on
It is a secret information restoration program of a secret information restoration device that realizes a secret data reproduction function for extracting ν pieces of secret data m concealed in the stego object s.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The sum of the contributions satisfies a ₁ + a ₂ +... + A _n = 1, and the scaling factor A is a positive real number less than 1.

  By using the secret information restoration apparatus and secret information restoration program according to the present invention, morphing is performed from morphing data that conceals different types of secret elements expressed by numerical values of the same format (for example, unsigned 8-bit integers). Secret data can be extracted (restored or reproduced) using steganography based on the above.

Furthermore, the secret information concealment device according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Data {m ₁ , m ₂ ,..., M _n } expressed as numerical values as the secret data m is set, and reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c. The feature vector {F ₁ , F ₂ ,..., F _n } is set as the stego key k _s , the stego object s to be generated is the morphing data I _0, and the feature vector of the morphing data I ₀ is F _0. And the scaling factor is A,
A scaling factor calculation means for obtaining a scaling factor A based on
a _i = m _i × A (values 1 to n are sequentially substituted for i)
, The contribution degree a _i (i = 1, 2,..., N) of each reference data I _i (i = 1, 2,..., N) is converted to the secret data m _i (i = 1, 2, n). .., N) a contribution degree calculation means determined based on
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on
By obtaining the morphing data I ₀ based on
Morphing data generation means for generating morphing data that conceals data expressed numerically with respect to reference data.

Moreover, the secret information concealment program of the secret information concealment apparatus according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information concealment program of a secret information concealment device for realizing a method of generating a stego object s in which secret data is concealed,
Data {m ₁ , m ₂ ,..., M _n } expressed as numerical values as the secret data m is set, and reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c. The feature vector {F ₁ , F ₂ ,..., F _n } is set as the stego key k _s , the stego object s to be generated is the morphing data I _0, and the feature vector of the morphing data I ₀ is F _0. And the scaling factor is A,
In the calculation means of the secret information concealment device,
A scaling factor calculation function for obtaining a scaling factor A based on
a _i = m _i × A (values 1 to n are sequentially substituted for i)
, The contribution degree a _i (i = 1, 2,..., N) of each reference data I _i (i = 1, 2,..., N) is converted to the secret data m _i (i = 1, 2, n). .., N) a contribution calculation function to be determined based on
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on ,
By obtaining the morphing data I ₀ based on
It is a secret information concealment program of a secret information concealment device that realizes a morphing data generation function for generating morphing data that conceals data expressed numerically with respect to reference data.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Also, I 0p, I w 1p,}} ···, I w np , _{respectively,} shown _{^{I 0, I w 1, ···}} , a p-th element of ^I _{w n.} Further, the scaling factor A is a positive real number less than 1.

By using the secret information concealment device and the secret information concealment program according to the present invention, a plurality of secrets expressed numerically in a plurality of cover data {I ₁ , I ₂ , I ₃ ,..., I _n } Data {m ₁ , m ₂ ,..., M _n } can be concealed. Also, the secret data need not be morphable.

Furthermore, the secret information restoration apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information restoration device for restoring secret data from a stego object s,
Set the morphing data _{I 0} as stego object s, cover reference data as the data _{c {I 1, I 2,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } are set, and the extracted secret data m is set to n pieces of data {m ₁ , m ₂ ,... M _n } represented by numerical values, and the feature vector of the morphing data I ₀ Is F ₀ and the scaling factor is A,
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on
When
Based on these two equations, the solution of the simultaneous equations is obtained, and the contribution a _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N). ) For calculating the degree of contribution
Based on the calculated contribution a _j (j = 1, 2,..., N),
To extract the secret data m _j (j = 1, 2,..., N),
It has a secret data reproduction means for extracting secret data hidden in the reference data from the morphing data.

Further, the secret information restoring program of the secret information restoring apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information restoration program of a secret information restoration device for realizing a method of restoring secret data from a stego object s,
Set the morphing data _{I 0} as stego object s, cover reference data as the data _{c {I 1, I 2,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } are set, and the extracted secret data m is set to n pieces of data {m ₁ , m ₂ ,... M _n } represented by numerical values, and the feature vector of the morphing data I ₀ Is F ₀ and the scaling factor is A,
In the calculation means of the secret information restoration device
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on ,
When
Based on these two equations, the solution of the simultaneous equations is obtained, and the contribution a _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N). )
Based on the calculated contribution a _j (j = 1, 2,..., N),
To extract the secret data m _j (j = 1, 2,..., N),
It is a secret information restoration program of a secret information restoration device for realizing a secret data reproduction function for extracting secret data concealed in reference data from morphing data.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Also, I 0p, I w 1p,}} ···, I w np , _{respectively,} shown _{^{I 0, I w 1, ···}} , a p-th element of ^I _{w n.} Further, the scaling factor A is a positive real number less than 1.

  By using the secret information restoration apparatus and the secret information restoration program according to the present invention, data represented by numerical values concealed in reference data can be extracted from the morphing data. Moreover, since the extracted data is secret data concealed using a plurality of cover data, the security of the secret data can be improved.

Furthermore, the secret information concealment device according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Set the reference data _{I 1} the secret data m, the cover reference data as the data _{c {I 2, I 3,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the generated stego object s is set as morphing data I ₀ .
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for transforming the secret data and the reference data to obtain I ^w _j (j = 1, 2,..., N) based on
By obtaining the morphing data (stego object) I ₀ based on
I ₀ generating means for concealing secret data from a plurality (three or more) of reference data.

Moreover, the secret information concealment program of the secret information concealment apparatus according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information concealment program of a secret information concealment device for realizing a method of generating a stego object s in which secret data is concealed,
Set the reference data _{I 1} the secret data m, the cover reference data as the data _{c {I 2, I 3,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the generated stego object s is set as morphing data I ₀ .
In the calculation means of the secret information concealment device
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
And a data transformation function for transforming the secret data and the reference data to obtain I ^w _j (j = 1, 2,..., N),
By determining the morphing data (stego object) I ₀ based on
It is a secret information concealment program of a secret information concealment device for realizing an I ₀ generation function for concealing secret data from a plurality (three or more) of reference data.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Also, I 0p, I w 1p,}} ···, I w np , _{respectively,} shown _{^{I 0, I w 1, ···}} , a p-th element of ^I _{w n.} Furthermore, the value of the sum of contributions shall satisfy a ₁ + a ₂ +... + _An = 1.

By using the secret information concealment device and the secret information concealment program according to the present invention, the secret data I ₁ is concealed in a plurality (n−1) of cover data {I ₂ , I ₃ ,..., I _n }. It becomes possible to do.

Furthermore, the secret information restoration apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information restoration device for restoring secret data from a stego object s,
Morphing data I ₀ is set as a stego object s, reference data {I ₂ , I ₃ ,..., I _n } is set as cover data c, and feature vectors {F ₁ , F ₂ are set as stego keys k _s. ,..., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the extracted secret data m is set as reference data I ₁ .
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for obtaining transformed data I ^w _j (j = 2, 3,..., N) of each reference data I _j (j = 2, 3,..., N) based on
To obtain the modified data I ^w ₁ of the secret data I ₁ ,
By extracting the secret data I ₁ based on
And secret data reproduction means for extracting secret data concealed with respect to a plurality of reference data.

Further, the secret information restoring program of the secret information restoring apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information restoration program of a secret information restoration device for realizing a method of restoring secret data from a stego object s,
Morphing data I ₀ is set as a stego object s, reference data {I ₂ , I ₃ ,..., I _n } is set as cover data c, and feature vectors {F ₁ , F ₂ are set as stego keys k _s. ,..., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the extracted secret data m is set as reference data I ₁ .
In the calculation means of the secret information restoration device,
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
A data transformation function for obtaining the transformed data I ^w _j (j = 2, 3,..., N) of each reference data I _j (j = 2, 3,..., N) based on ,
To obtain the modified data I ^w ₁ of the secret data I ₁ ,
By extracting the secret data I ₁ based on
A secret information restoration program of a secret information restoration device for realizing a secret data reproduction function for extracting secret data concealed with respect to a plurality of reference data.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Also, I 0p, I w 1p,}} ···, I w np , _{respectively,} shown _{^{I 0, I w 1, ···}} , a p-th element of ^I _{w n.} Furthermore, the value of the sum of contributions shall satisfy a ₁ + a ₂ +... + _An = 1.

By using the secret information restoration device and the secret information restoration program according to the present invention, it is possible to extract (reproduce) the secret data I ₁ from the morphing data I ₀ . Here, since the secret data I ₁ is data concealed by a plurality (n−1) of reference data {I ₂ , I ₃ ,..., I _n }, the morphing data I ₀ is a plurality of reference data. Is the transformed (morphed) data. For this reason, it is possible to improve the security of the concealed secret data as compared with the case of performing steganography based on morphing using two reference data.

Also, the secret information concealment device according to the present invention is
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
s _p = MSB (I ⁰ _0p ) + m _q
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
and a stego object generating means for generating a stego object s in which v secret data is concealed.

Moreover, the secret information concealment program of the secret information concealment apparatus according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment program of a secret information concealment device for realizing a method of generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
s _p = MSB (I ⁰ _0p ) + m _q
With seeking stego object s _p, the value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
It is a secret information concealment program of a secret information concealment device for realizing a stego object generation function for generating a stego object s in which v secret data is concealed.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The value of the sum of contributions satisfies a ₁ + a ₂ +... + A _n = 1. The secret data m _q (q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and the MSB (I ⁰ _0p ) is I ^0. _This means a function for extracting the upper _Bb bit data from _0p .

  By using the secret information concealment device and the secret information concealment program according to the present invention, the secret data is not concealed in the morphing data along with the morphing process, but is generated after the morphing data is generated by the morphing process. Since the secret data is concealed, the secret data itself is not subject to the morphing process. For this reason, the stego object in which the secret data is concealed is less likely to be unnatural even if the secret data is concealed. Therefore, even if a third party sees the stego object, it becomes difficult to notice the concealment of the secret data, and the possibility of being the target of the attack can be reduced.

By using the secret information concealment apparatus and the secret information concealment program according to the present invention, the secret data is concealed in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. It becomes possible. For this reason, it is difficult for a third party to easily determine at which position the secret data is concealed, and the security of the secret data can be improved.

Moreover, the secret information restoring apparatus according to the present invention is
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
_{Δ p (j 1, j 2} ) = (I w j1, p -I w j2, p) if and ≠ 0 become _{j 1} and _{j 2} are present, m based on the corresponding _{j 1} and _{j 2 q} = s _p -MSB (s _p )
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
And secret data reproduction means for extracting ν secret data concealed in the stego object.

Further, the secret information restoring program of the secret information restoring apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration program of a secret information restoration device for realizing a method of restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
_{Δ p (j 1, j 2} ) = (I w j1, p -I w j2, p) if and ≠ 0 become _{j 1} and _{j 2} are present, m based on the corresponding _{j 1} and _{j 2 q} = s _p -MSB (s _p )
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining the above and repeatedly executing the determination process until q ≧ ν,
It is a secret information restoration program of a secret information restoration device for realizing a secret data reproduction function for extracting v secret data concealed in a stego object.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The value of the sum of contributions satisfies a ₁ + a ₂ +... + A _n = 1. Further, secret data _{m q (q = 1,2, ···} , ν) is the data of b ^bits, _{I 0 0p} is a B data bits, MSB _{(s p)} is the _{s p} Of these, it means a function for taking out the data of upper Bb bits.

By using the secret information restoration device and the secret information restoration program according to the present invention, the secret data is extracted from the stego object s in which a plurality of secret data is concealed by obtaining the concealed location of the secret data. It becomes possible. Specifically, I ^w _{j1, p} and I ^w _{j2, p} are obtained using the cover data and the stego key, and Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, Since} it is necessary to obtain p where _{p 1} ) ≠ 0, even if only the stego object s is acquired, the secret data m cannot be extracted, and the security of the secret data m can be improved.

Also, the secret information concealment device according to the present invention is
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
and a stego object generating means for generating a stego object s in which v secret data is concealed.

Moreover, the secret information concealment program of the secret information concealment apparatus according to the present invention is:
For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment program of a secret information concealment device for realizing a method of generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Let me seek stego object s _p with a value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
It is a secret information concealment program of a secret information concealment device for realizing a stego object generation function for generating a stego object s in which v secret data is concealed.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The value of the sum of contributions satisfies a ₁ + a ₂ +... + A _n = 1. The secret data m _q (q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and the MSB (I ⁰ _0p ) is I ^0. _This means a function for extracting the upper _Bb bit data from _0p . further,
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits Data consisting of numerical values is set.

By using the secret information concealment apparatus and the secret information concealment program according to the present invention, the secret data is concealed in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. It becomes possible. Furthermore, the secret data to be concealed is concealed in a state where the exclusive OR operation is performed with the b-bit value obtained from the secret data and the morphing data or the like. For this reason, it is possible to make it difficult for a third party to determine that the secret data is concealed, and it is possible to reduce the risk of the secret data being extracted by the third party.

Moreover, the secret information restoring apparatus according to the present invention is
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
And secret data reproduction means for extracting ν secret data concealed in the stego object.

Further, the secret information restoring program of the secret information restoring apparatus according to the present invention is:
For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration program of a secret information restoration device for realizing a method of restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining the above and repeatedly executing the determination process until q ≧ ν,
It is a secret information restoration program of a secret information restoration device for realizing a secret data reproduction function for extracting v secret data concealed in a stego object.

However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... , n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector _{F 0} Have ^{_{Further, s p, I 0 0p,}} I w 1p, ···, I w np , ^{respectively,} show ^{_{s, I 0 0, I w}} 1, ···, a p-th element of ^I _{w n.} Further, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the values of the contributions a _j (j = 1, 2,..., N) are normalized in advance, The value of the sum of contributions satisfies a ₁ + a ₂ +... + A _n = 1. Further, secret data _{m q (q = 1,2, ···} , ν) is the data of b ^bits, _{I 0 0p} is a B data bits, MSB _{(s p)} is the _{s p} Of these, it means a function for taking out the data of upper Bb bits. further,
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits Data consisting of numerical values is set.

  By using the secret information restoration device and the secret information restoration program according to the present invention, the secret data is concealed from the stego object s in which a plurality of secret data is concealed, and the data is extracted. Since the secret data is extracted by performing an exclusive OR operation on the extracted data and the value of y, security can be improved. In addition, it is possible to ensure sufficient security without encryption, and it is possible to improve the extraction accuracy because secret data can be accurately reproduced if there is no communication error. It becomes.

  By using the secret information concealment apparatus and secret information concealment program according to the present invention, a stegano based on morphing can be used even if the number of secret data ν is relatively large as long as it is within the range of the number of elements of reference data. It is possible to conceal the secret data from the reference data by using the graphic.

Further, by using the secret information concealment apparatus and the secret information concealment program according to the present invention, secret data {m ₁ that cannot be morphed into a plurality of cover data {I ₁ , I ₂ ,..., I _n }. , M ₂ ,..., M _n } can be concealed.

Further, by using the secret information hiding apparatus and secret information concealed program according to the present invention, a plurality (n-1 pieces) cover data _{{I 2, I 3, ···} , I n} , the secret data _{I 1} Can be concealed.

  On the other hand, by using the secret information restoration apparatus and the secret information restoration program according to the present invention, the secret data is extracted (restored, using morphing-based steganography from the morphing data in which ν pieces of secret data are concealed. Reproducible).

  In addition, by using the secret information restoration apparatus and the secret information restoration program according to the present invention, it is possible to extract from the morphing data data that is hidden in the reference data and cannot be morphed. Furthermore, since the extracted data is secret data concealed using a plurality of cover data, the security of the secret data can be improved.

Furthermore, by using the secret information restoration apparatus and secret information restoration program according to the present invention, it is possible to extract (reproduce) the secret data I ₁ from the morphing data I ₀ . Here, since the secret data I ₁ is data concealed by a plurality (n−1) of reference data {I ₂ , I ₃ ,..., I _n }, the morphing data I ₀ is a plurality of reference data. Is the transformed (morphed) data. For this reason, it is possible to improve the security of the concealed secret data as compared with the case of performing steganography based on morphing using two reference data.

  Also, by using the secret information concealment device and the secret information concealment program according to the present invention, the secret data is not concealed in the morphing data by the morphing process, but after the morphing data is generated by the morphing process, the morphing Since the secret data is concealed in the data, the secret data itself is not subject to the morphing process. For this reason, the stego object in which the secret data is concealed is less likely to be unnatural even if the secret data is concealed. Therefore, even if a third party sees the stego object, it becomes difficult to notice the concealment of the secret data, and the possibility of being the target of the attack can be reduced.

On the other hand, by using the secret information concealment apparatus and the secret information concealment program according to the present invention, the secret data is stored in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. Can be concealed. For this reason, it is difficult for a third party to easily determine at which position the secret data is concealed, and the security of the secret data can be improved.

Further, by using the secret information restoration apparatus and secret information restoration program according to the present invention, the secret data is extracted by obtaining the hidden location of the secret data from the stego object s where the secret data is hidden. Can be performed. Specifically, I ^w _{j1, p} and I ^w _{j2, p} are obtained using the cover data and the stego key, and Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, Since} it is necessary to obtain p where _{p 1} ) ≠ 0, even if only the stego object s is acquired, the secret data m cannot be extracted, and the security of the secret data m can be improved.

Furthermore, by using the secret information concealment device and the secret information concealment program according to the present invention, the secret data is stored in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. It is possible to hide. Furthermore, the secret data to be concealed is concealed in a state where the exclusive OR operation is performed with the b-bit value obtained from the secret data and the morphing data or the like. For this reason, it is possible to make it difficult for a third party to determine that the secret data is concealed, and it is possible to reduce the risk of the secret data being extracted by the third party.

  In addition, by using the secret information restoration apparatus and the secret information restoration program according to the present invention, the secret data is concealed from the stego object s where the secret data is concealed, and the data is extracted. In addition, since the secret data is extracted by performing an exclusive OR operation on the extracted data and the value of y, security can be improved. In addition, it is possible to ensure sufficient security without encryption, and it is possible to improve the extraction accuracy because secret data can be accurately reproduced if there is no communication error. It becomes.

It is the schematic block diagram which showed an example of the state with which the secret information concealment apparatus and the secret information restoration apparatus were connected via the open network. It is the block diagram which showed schematic structure of the secret information concealment apparatus which concerns on embodiment. 6 is a functional block diagram when a calculation unit according to Embodiment 1 reads a program stored in a ROM and performs a concealment process of secret data. FIG. (A) is the flowchart which showed the process which the calculation part which concerns on Embodiment 1 conceals secret data, (b) is the process which the calculation part which concerns on Embodiment 1 extracts secret data. It is the shown flowchart. FIG. 5 is a diagram illustrating functional blocks when the calculation unit according to the first embodiment reads a program stored in a ROM and performs a secret data extraction process. FIG. 10 is a diagram illustrating a functional block when a calculation unit according to a second embodiment reads a program stored in a ROM and performs a secret data concealment process. (A) is the flowchart which showed the process in which the calculation part which concerns on Embodiment 2 conceals secret data, (b) is the process in which the calculation part which concerns on Embodiment 2 extracts secret data. It is the shown flowchart. It is a figure showing a functional block in case a calculation part concerning Embodiment 2 reads a program memorized by ROM, and performs extraction processing of secret data. FIG. 10 is a diagram illustrating functional blocks when a calculation unit according to a third embodiment reads a program stored in a ROM and performs a secret data concealment process. 10 is a flowchart illustrating a process in which a calculation unit according to Embodiment 3 reads a program stored in a ROM and conceals secret data. FIG. 10 is a functional block diagram when a calculation unit according to a third embodiment reads a program stored in a ROM and performs a secret data extraction process. 12 is a flowchart illustrating a process in which a calculation unit according to Embodiment 3 reads a program stored in a ROM and extracts secret data. Using the secret information concealment device described in the third embodiment, the secret image is concealed in the morphing image when the product of the secret data x _q and (I ^w _{j1, p} −I ^w _{j2, p} ) is a little larger (A) is two reference images, (b) is a morphing image, (c) is a stego image in which a secret image is concealed in the morphing image, and (d) is before concealment. (E) is a secret image extracted from the stego image. 10 is a flowchart illustrating processing in which a calculation unit according to Embodiment 4 reads a program stored in a ROM and conceals secret data. FIG. 10 is a diagram illustrating functional blocks when a calculation unit according to a fourth embodiment reads a program stored in a ROM and performs secret data extraction processing. 14 is a flowchart illustrating processing in which a calculation unit according to Embodiment 4 reads out a program stored in a ROM and extracts secret data. FIG. 6 is an example of an image when a secret image is concealed in a morphing image using the secret information concealment device described in Embodiment 4, wherein (a) is a stego image in which the secret image is concealed and (b) is concealed (C) is a secret image extracted from the stego image. 10 is a flowchart illustrating a process in which a calculation unit according to Embodiment 5 reads a program stored in a ROM and conceals secret data. FIG. 10 is a diagram illustrating a functional block when a calculation unit according to a fifth embodiment reads a program stored in a ROM and performs a secret data extraction process. 10 is a flowchart illustrating processing in which a calculation unit according to Embodiment 5 reads out a program stored in a ROM and extracts secret data. 14 is a flowchart illustrating processing in which a calculation unit according to Embodiment 6 reads a program stored in a ROM and conceals secret data. 14 is a flowchart illustrating a process in which a calculation unit according to Embodiment 6 reads a program stored in a ROM and extracts secret data. It is the schematic block diagram which showed an example of the authentication system of a card user. It is an image example in the case of concealing an image using the confidential information concealment device described in Embodiment 3, (a) is a cover image, (b) is a secret image, (c) is a morphing image, and (d) is an image. It is an extracted image extracted from the morphing image. It is the figure which showed an example of the sequence of the preparatory work before performing an authentication process between a server, a local terminal, and a user's portable device. It is the figure which showed an example of the sequence in the case of performing an authentication process between a server, a local terminal, and a user's portable device.

  Hereinafter, a secret information concealment device and a secret information restoration device according to the present invention will be described in detail with reference to an example. FIG. 1 shows a schematic configuration in which a computer A of a user A and a computer B of a user B are connected to a server 4 via a network 3. In the computer A, the morphing data can be generated by concealing the secret data in the cover data. By transmitting the morphing data from the computer A to the computer B of the user B via the network 3, the computer B It is possible to extract (restore and reproduce) secret data from morphing data.

  Here, since the computer A of the user A is a computer used to conceal the secret data from the cover data, it corresponds to the secret information concealment device 1. The computer B of the user B corresponds to the secret information restoring device 2 because it is a computer used to extract (restore) secret data from the morphing data. However, since the secret data concealment process and the extraction process can generally be realized by reversing the same process, both the computer A of the user A and the computer B of the user B are protected by the secret data. It may be provided with a function as a secret information concealment / restoration apparatus capable of performing the concealment process and the restoration process.

  The computer A and the computer B are connected via an open network 3 such as the Internet. For convenience of explanation, FIG. 1 shows only the computer A as the secret information concealment device 1 and the computer B as the secret information restoration device 2, but there are many computers other than the computer A and the computer B. It may be connected via the network 3. Each computer connected via the network 3 can send and acquire information directly or indirectly via the web server 4 or mail server.

  Furthermore, it is possible to send and receive data from computer A to computer B using a communication method different from that of the open network 3. Unlike the network 3, this communication method employs a communication method in which data cannot be confirmed by users other than the computer A and the computer B. For example, a communication method using a dedicated line, a method of storing data in a storage medium such as a CD or USB, and mailing the data can be used as an example. Hereinafter, this communication method is referred to as a closed network communication method in order to distinguish it from an open network communication method.

  FIG. 2 is a block diagram showing a schematic configuration of the secret information concealment apparatus 1. The secret information concealment device 1 can be configured by a general computer (computer, mobile phone, etc.). The device main body 6, a display unit 7, a data input unit 8, a command input unit 9, have.

  The display unit 7 corresponds to a general liquid crystal display or a CRT display (CRT display), and has a function of displaying processing contents and the like so as to be visible to the user.

  The data input unit 8 and the command input unit 9 are input means for the user to input information, commands, and the like necessary for processing of the secret information concealment device 1. Specifically, the data input unit 8 has a function of inputting cover data, secret data, and the like. For example, when cover data, stego objects, secret data, and the like are image data, a digital camera, a scanner, and the like correspond to examples. Further, a USB terminal, a CD drive, a DVD drive, and the like for taking cover data, stego objects, secret data, stego keys, and the like as data into a computer correspond to examples.

  Furthermore, the command input unit 9 has a function of inputting execution commands, correction commands, and the like, and corresponds to a keyboard, a mouse, a touch pen, and the like.

  The apparatus body 6 includes an algorithm execution unit 10, a communication unit 11, and an auxiliary storage unit 12. The algorithm execution unit 10 has a role of mainly performing morphing data generation processing, secret data concealment processing, and the like.

  The algorithm execution unit 10 includes a calculation unit (calculation unit) 15 and a storage unit 18.

  The calculation unit 15 functions as a calculation unit according to the present invention, and has a role of performing various calculation / control processes in the secret information concealment apparatus 1 such as a process of generating a stego object s. The storage unit 18 has a function of temporarily storing data necessary for processing in the calculation unit 15 and calculation results (for example, modified data).

  The storage unit 18 includes, for example, a general ROM (Read Only Memory) 16 and a RAM (Random Access Memory) 17. The ROM 16 stores a processing program or the like when the secret information concealment device 1 is activated. When the secret information concealment device 1 is activated or the like, the calculation unit 15 reads the processing program or the like in the ROM 16 to perform activation processing or the like. (Realization of methods such as startup processing) is possible. The RAM 17 is used as a work area for processing performed in the calculation unit 15.

  The communication unit 11 corresponds to a network interface card (NIC), for example, and has a role of connecting the secret information concealment device 1 to the network 3. In FIG. 2, for convenience, the network card built in the apparatus main body 6 is described as the communication unit 11. May be connected. In this case, the communication unit 11 includes a modem or the like.

  The auxiliary storage unit 12 is a so-called auxiliary storage device, and is mainly composed of a hard disk. The auxiliary storage unit 12 can store data necessary for processing of the secret information concealment device 1, such as secret data, cover data, stego object, and stego key information.

  The auxiliary storage unit 12 can store data necessary for the secret information concealment process (for example, modified data generated by the secret information concealment device 1, morphing data, etc.). The auxiliary storage unit 12 is not limited to only a hard disk, but is a storage unit (for example, an SSD (Solid State Drive)) configured by a flash memory or the like, or a storage unit using a tape medium (for example, DAT (Digital Audio) Tape)) etc.

  The calculation unit 15 executes a stego object s generation process (generation process: concealment process / concealment method for the secret data m with respect to the cover data c) in accordance with a processing program stored in the auxiliary storage unit 12 or the ROM 16. In FIG. 2, only one calculation unit 15 is provided as an arithmetic processing circuit, but a plurality of arithmetic processing circuits may be provided according to each function.

  On the other hand, as described above, the secret data concealment process and the extraction process can generally be realized by performing the same process in reverse. Therefore, both the computer A of the user A and the computer B of the user B may have a function as a secret information concealment / restoration device that can perform the concealment process and the restoration process of the secret data, The configuration of the secret information restoring device 2 also has the same configuration as the block diagram of the schematic configuration shown in FIG. For this reason, the secret information restoration device 2 is configured by the same functional units as the secret information concealment device 1, and the same reference numerals are used to indicate the functional blocks, and detailed description thereof is omitted.

  In the embodiment, it is assumed that each computer performs both processing of concealing secret data and processing of restoration. Further, whether the computer implements the secret information concealing function or the secret information restoration function, the processing is performed based on the functional blocks shown in FIG. For this reason, when functioning as the secret information restoration device 2, the calculation unit 15 executes a process of extracting secret data from the stego object according to a program for realizing a processing method stored in the auxiliary storage unit 12 or the ROM 16. To do.

  Next, the secret data concealment process and extraction process described above will be described in detail by dividing them into a plurality of embodiments.

[Embodiment 1]
First, as Embodiment 1, the calculation unit 15 generates a stego object that conceals a plurality of secret data with respect to a plurality of reference data, and extracts secret data from the stego object generated in this way. The processing will be described.

  FIG. 3 is a diagram illustrating functional blocks when the calculation unit 15 reads a program stored in the ROM 16 and performs a concealment process of secret data (implementing a concealment process method) in the first embodiment.

In FIG. 3, a data transformation unit (data transformation unit) 21, an F ₀ generation unit (F ₀ generation unit) 22, and an I ₀ generation unit 23 (I ₀ generation unit) perform concealment processing according to a program (concealment). The calculation unit 15 (which realizes the processing method) substantially corresponds. Further, the secret data storage unit 24, the cover data storage unit 25, the stego key storage unit 26, the stego object storage unit 27, and the working memory 28 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Incidentally, the cover data storage unit 25, are stored in advance cover data c is the stego key storage unit 26, it is possible to assume that previously stego key k _s are stored. The secret data storage unit 24 stores secret data m in advance, and the stego object storage unit 27 stores the generated stego object s.

Further, the calculation unit 15 acquires the cover data c, the secret data m, or the stego key k _s through the open network 3 or the closed network, thereby acquiring the cover data acquisition unit, the secret data acquisition unit, or the stego key. It is also possible to have a configuration in which the acquired data is stored in the cover data storage unit 25, the secret data storage unit 24, or the stego key storage unit 26, respectively.

State, however, when using actual secret information hiding apparatus 1 shown in FIG. 3, the cover data c and the stego key k _s is previously stored in the cover data storage unit 25 and the stego key storage unit 26 It is thought that processing is often performed in

  Further, the calculation unit 15 functions as a stego object output unit as necessary, reads the stego object s from the stego object storage unit 27, and functions as the secret information restoration device 2 via the open network 3. It plays a role to output to.

  Next, the process of the calculation part 15 which concerns on Embodiment 1 is demonstrated according to the flowchart shown to Fig.4 (a).

  In the concealment process of the secret data according to the first embodiment, the secret data m is concealed in the cover data c using steganography using morphing based on 3 or more, for example, n (n ≧ 3) reference data Will be described.

When the secret data is concealed using n pieces of reference data, it can be defined as follows.
Secret data m = _{I 1,} the cover data _{c = {I 2, I 3} , ···, I n}, stego key _{k s = {F 1, F} 2, ···, F n; a 1, a 2 ,..., A _n }, stego object s = I ₀

_Here, I 1 ~I _n reference _data, F 1 to F _n, the feature vector corresponding to each of the reference data _{I 1 ~I n, a 1 ~a} n is the reference data _I 1 ~I _n each The degree of contribution corresponding to, I ₀ indicates morphing data. The contribution of each reference data is set independently, but in order to actually keep the elements of the feature vector and the elements of the morphing data (transformed data) within the specified dynamic range, the sum of the contributions must be Conditions for ₁ a ₁ + a ₂ +... + A _n = 1
It is necessary to satisfy. Therefore, it is only necessary to set n−1 contributions among the _n contributions a ₁ to an. Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of outputting the stego object s.

[Algorithm for realizing function f for concealing secret data]
_{Input: m = I 1, c =} {I 2, I 3, ···, I n}, k s = {F 1, F 2, ···, F n; a 1, a 2, ··· , A _n }
Output: s = I ₀
STEP 1 (Step S.1 shown in FIG. 4A)
The calculation unit 15 obtains a feature vector F ₀ of the morphing data I ₀ based on the following equation.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 22 shown in FIG.

STEP2 (Step S.2 shown in FIG. 4A)
Based on the following equation, the calculation unit 15 transforms the secret data and the reference data to obtain data I ^w _j (j = 1, 2,..., N).
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this processing, the calculation unit 15 functions as the data transformation unit 21 shown in FIG.

STEP 3 (Step S.3 shown in FIG. 4A)
The calculation unit 15 obtains morphing data (stego object) I ₀ based on the following equation.
^{_{However, I 0p, I w 1p,}} ···, I w np , ^{_{respectively, I 0, I w 1,}} ···, shows a p-th element of ^I _{w n.} In this process, the calculation unit 15 functions as the I ₀ generation unit 23 shown in FIG. The obtained morphing data (stego object) is stored in the stego object storage unit 27.

By thus through the STEP1~STEP3 procedure (the process shown in FIG. 4 (a)), the reference data of a plurality (n-1 _{piece) {I 2, I 3,} ···, I n} , the secret It is possible to obtain morphing data I ₀ in which data I ₁ is concealed.

On the other hand, the secret information restoration device 2 can extract secret data by performing a process of extracting (restoring and reproducing) the reference data I ₁ from the morphing data I ₀ generated by the concealment process described above. It becomes.

  FIG. 5 is a diagram showing functional blocks when the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing the extraction processing method).

In FIG. 5, a data transformation unit (data transformation unit) 31, an F ₀ generation unit (F ₀ generation unit) 32, and a secret data reproduction unit (secret data reproduction unit) 33 perform an extraction process according to a program (extraction). The calculation unit 15 (which realizes the processing method) substantially corresponds. Further, the secret data storage unit 34, the cover data storage unit 35, the stego key storage unit 36, the stego object storage unit 37, and the working memory 38 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

The stego object storage unit 37 stores the stego object s generated by the secret information concealment device 1 shown in FIG. 3, the cover data storage unit 35 stores the cover data c, and the stego key storage unit 36 stores the cover data c. is stored in the stego key k _s, the secret data storage unit 34, the secret data m is stored.

  In the first embodiment, the calculation unit 15 functions as a stego object acquisition unit, and the stego generated from the stego object storage unit 27 of the secret information concealment device 1 shown in FIG. 3 according to the process of FIG. The object s is acquired via the open network 3 and stored in the stego object storage unit 37. In addition, the calculation unit 15 functions as a cover data acquisition unit and a stego key acquisition unit, and acquires the cover data and the stego key used for the secret data concealment process in the secret information concealment device 1 via the closed network. , Processing to be stored in the cover data storage unit 35 and the stego key storage unit 36 is performed. It is also possible to adopt a configuration in which cover data and stego keys are stored in advance in the cover data storage unit 35 and the stego key storage unit 36 of all the secret information restoring devices 2.

  In addition, the calculation unit 15 functions as a secret data output unit as necessary, and reads secret data from the secret data storage unit 34 and displays the extracted data on the display unit 7 or the like.

  Next, the secret data extraction process of the calculation unit 15 in the first embodiment will be described with reference to the flowchart shown in FIG.

  In the secret data extraction processing according to the first embodiment, morphing is used using 3 or more, for example, n (n ≧ 3) reference data, as in the case described with reference to FIG. Based on steganography, a process of extracting secret data from the stego object is performed.

Therefore, when extracting secret data concealed using n pieces of reference data, it is defined as follows.
Stego object s = I ₀ , cover data c = {I ₂ , I ₃ ,..., I _n }, stego key k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a ₂ ,..., A _n }, secret data m = I ₁
There was thus defined, s, c, as input data to k _s, computing unit 15 performs a process of extracting secret data m.

[Algorithm for realizing function f- ¹ for extracting secret data]
_{Input: s = I 0, c =} {I 2, I 3, ···, I n}, k s = {F 1, F 2, ···, F n; a 1, a 2, ··· , A _n }
Output: m = I ₁
STEP 1 (Step S.11 shown in FIG. 4B)
The calculation unit 15 obtains a feature vector F ₀ of the morphing data I ₀ based on the following equation.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 32 shown in FIG.

STEP 2 (Step S.12 shown in FIG. 4B)
The calculation unit 15 uses the modified data I ^w _j (j = 2, 3,..., N) of each reference data I _j (j = 2, 3,..., N) based on the following equation. Ask.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 31 shown in FIG.

STEP 3 (Step S.13 shown in FIG. 4B)
The calculation unit 15 uses the modified data I ^w _j (j = 2, 3,..., N) to obtain a modified I ^w ₁ of the secret data I ₁ according to the following equation.
In this process, the calculation unit 15 functions as the secret data reproduction unit 33 shown in FIG. _{^{However, I 0p, I w 1p,}} ···, I w np , _{^{respectively, I 0, I w 1,}} ···, shows a p-th element of ^I _{w n.}

STEP 4 (Step S.14 shown in FIG. 4B)
The calculation unit 15 reproduces the secret data based on the following formula.
In this process, the calculation unit 15 functions as the secret data reproduction unit 33 shown in FIG. In addition, the secret data obtained by this process is stored in the secret data storage unit 34.

As described above, the secret data I ₁ can be extracted (reproduced) from the morphing data I ₀ through the steps 1 to 4 described above (the process shown in FIG. 4B). Here, since the secret data I ₁ is data concealed by a plurality (n−1) of reference data {I ₂ , I ₃ ,..., I _n }, the morphing data I ₀ is a plurality of reference data. Is the transformed (morphed) data.

  For this reason, compared with the case where steganography based on morphing is performed using two reference data, the security of the hidden secret data can be improved.

[Embodiment 2]
Next, as Embodiment 2, a process of concealing and extracting secret data from a plurality of reference data when using data that cannot be morphed as secret data will be described.

  FIG. 6 is a diagram illustrating functional blocks when the calculation unit 15 reads a program stored in the ROM 16 and performs a concealment process of secret data (realizes a concealment process method) in the second embodiment.

In FIG. 6, a scaling factor calculation unit (scaling factor calculation unit) 41, a contribution degree calculation unit (contribution degree calculation unit) 42, a data transformation unit (data transformation unit) 43, and an F ₀ generation unit (F ₀ generation unit). ) 44 and the morphing data generation unit (morphing data generation means) 46 substantially correspond to the calculation unit 15 that performs concealment processing (implementing the concealment processing method) according to a program. Further, the secret data storage unit 47, the cover data storage unit 48, the stego key storage unit 49, the stego object storage unit 50, and the working memory 51 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Further, the cover data storage unit 48, are stored in advance cover data c is the stego key storage unit 49, it is possible to assume that previously stego key k _s are stored. The secret data storage unit 47 stores secret data m in advance, and the stego object storage unit 50 stores the generated stego object s.

  The calculation unit 15 acquires the cover data, the secret data, or the stego key used for the secret data concealment process via the open network 3 or the closed network, so that the cover data acquisition unit, the secret data acquisition unit Alternatively, it may be configured to function as a stego key acquisition unit and store the acquired data in the cover data storage unit 48, the secret data storage unit 47, or the stego key storage unit 49, respectively.

  However, when the secret information concealment device 1 shown in FIG. 6 is actually used, cover data and a stego key may be acquired in advance and stored in the cover data storage unit 48 and the stego key storage unit 49. It is thought that there are many.

  In addition, the calculation unit 15 functions as a stego object output unit as needed, reads a stego object from the stego object storage unit 50, and transmits it to the computer functioning as the secret information restoration device 2 via the open network 3. It plays the role to output.

  Next, the process of the calculation part 15 which concerns on Embodiment 2 is demonstrated according to the flowchart shown to Fig.7 (a).

As described above, in the first embodiment, it has been described the case of using the reference data I ₁ is morphable data as confidential data. As an example of morphable data, biometric information such as a face image, a fingerprint, a vein pattern, and voice can be used. By performing data concealment processing using these biometric information as secret data, the secret information concealment device 1 and the secret information shown in the first embodiment for the customer authentication processing specified by the face image, fingerprint, vein pattern, voice, etc. The restoration device 2 can be used. Furthermore, it is possible to secure higher security than the conventional authentication technology and improve the reliability of the authentication system.

  However, in an actual authentication system, morphable data and non-morphable data may be simultaneously used as secret data. For example, data represented by numerical values such as a member number, a telephone number, an address, a purchase history, and an action pattern correspond to data that cannot be morphed. If data that cannot be morphed by being expressed numerically in this way can also be concealed by steganography based on morphing, the utility of the authentication system can be further enhanced.

  In the secret data concealment process according to the second embodiment, a process for concealing secret data in cover data when data that cannot be morphed (represented by numerical values) is used as the secret data will be described.

In the secret data concealment process according to the second embodiment, the definition is as follows.
Secret data _{m = {m 1, m 2} , ···, m n}, the cover data _{c = {I 1, I 2} , ···, I n}, stego key _{k s = {F 1, F} 2, ..., F _n }, stego object s = {I ₀ , F ₀ , A}
_Here, m 1 ~m _n is secret data consisting unable morphing _data, I 1 ~I _n reference _data, F 1 to F _n is the feature vector corresponding to each of the reference data _I 1 ~I _{n, I 0} And F ₀ indicate morphing data obtained from the reference data and feature vectors of the morphing data.

A is a scaling factor for obtaining a contribution {a ₁ , a ₂ ,..., A _n } that satisfies the condition of a ₁ + a ₂ + a ₃ +... + A _n = 1. This scaling factor can be obtained by the following equation.
By setting the cover data c and the stego key k _s, can be embedded secret data m to the stego object s. Here, it is assumed that all secret data is expressed by numerical values.

Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of obtaining the stego object s (generation).

[Algorithm for realizing function f for concealing secret data]
Input: m = {m ₁ , m ₂ ,..., M _n }, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,. F _n }
Output: s = {I ₀ , F ₀ , A}

STEP 1 (Step S.21 shown in FIG. 7A)
The calculation part 15 calculates | requires the scaling factor A based on following Formula.
In this process, the calculation unit 15 functions as a scaling factor calculation unit 41 shown in FIG.

STEP 2 (Step S.22 shown in FIG. 7A)
The calculating unit 15 calculates the contribution a _i (i = 1, 2,..., N) of each reference data I _i (i = 1, 2 _,. Determine based on (i ₁ = 1, 2,..., N).
a _i = m _i × A (values 1 to n are sequentially substituted for i)
In this process, the calculation unit 15 functions as the contribution calculation unit 42 shown in FIG.

STEP 3 (Step S.23 shown in FIG. 7A)
The calculation unit 15 obtains a feature vector F ₀ of the morphing data I ₀ based on the following equation.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 44 illustrated in FIG.

STEP 4 (Step S.24 shown in FIG. 7A)
Based on the following equation, the calculation unit 15 _generates the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N). Ask.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 43 shown in FIG.

STEP 5 (Step S.25 shown in FIG. 7A)
The calculation unit 15 obtains morphing data (stego object) I ₀ based on the following equation.
^{_{However, I 0p, I w 1p,}} ···, I w np , ^{_{respectively, I 0, I w 1,}} ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as the morphing data generation unit 46 shown in FIG. The obtained morphing data is stored in the stego object storage unit 50.

In this way, through the procedure of STEP 1 to STEP 5 (the process shown in FIG. 7A), it is not possible to morph the plurality of cover data {I ₁ , I ₂ ,..., I _n } (for example, numerical values). it is possible to obtain the morphing data _{I 0} that representation has been) a plurality of secret data _{{m 1, m 2, ···} , m n} is concealed. Moreover, even if it is a plurality of secret data expressed by numerical values, the cover data can be concealed, and therefore, the secret data does not necessarily need to be morphable data.

On the other hand, the secret information restoration device 2 extracts (restores) a plurality of secret data {m ₁ , m ₂ ,..., M _n } from the morphing data I ₀ generated by the concealment process described above. Can be done.

  FIG. 8 is a diagram showing functional blocks when the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing an extraction processing method).

In FIG. 8, a data transformation unit (data transformation unit) 61, a contribution calculation unit (contribution calculation unit) 62, and a secret data reproduction unit (secret data reproduction unit) 63 perform extraction processing according to a program (extraction). The calculation unit 15 (which realizes the processing method) substantially corresponds. The secret data storage unit 64, the cover data storage unit 65, the stego key storage unit 66, the stego object storage unit 67, and the working memory 68 substantially correspond to the RAM 17 or the auxiliary storage unit 12. The stego object storage unit 67 stores a stego object s generated by the secret information concealment device 1 shown in FIG. 6, the cover data storage unit 65 stores cover data c, and the stego key storage unit 66 stores , The stego key k _s is stored, and the secret data storage unit 64 stores the secret data m.

  In the second embodiment, the calculation unit 15 functions as a stego object acquisition unit, and is generated from the stego object storage unit 50 of the secret information concealment device 1 shown in FIG. 6 by the process shown in FIG. Stego objects (morphing data) are acquired via the open network 3 and stored in the stego object storage unit 67. Further, the calculation unit 15 functions as a cover data acquisition unit and a stego key acquisition unit, and acquires the cover data and the stego key used for the secret data concealment process in the secret information concealment device 1 via the closed network. The cover data storage unit 65 and the stego key storage unit 66 may be stored.

  It is also possible to adopt a configuration in which cover data and stego keys are stored in advance in the cover data storage unit 65 and the stego key storage unit 66 of all the secret information restoring devices 2. Further, the calculation unit 15 functions as a secret data output unit as necessary, and reads secret data from the secret data storage unit 64 and displays the extracted data on the display unit 7 or the like.

  Next, the secret data extraction process of the calculation unit 15 in the second embodiment will be described with reference to the flowchart shown in FIG. In the secret data extraction process according to the second embodiment, a case will be described in which secret data is extracted from a stego object in which secret data that cannot be morphed (for example, expressed by a numerical value) is concealed.

When extracting n secret data concealed using n reference data, it is defined as follows.
Stego object s = {I ₀ , F ₀ , A}, cover data c = {I ₁ , I ₂ ,..., I _n }, stego key k _s = {F ₁ , F ₂ ,. _n }, secret data m = {m ₁ , m ₂ ,..., m _n }
There was thus defined, s, c, as input data to k _s, computing unit 15 performs a process of extracting secret data m.

[Algorithm for realizing function f- ¹ for extracting secret data]
_{Input: s = {I 0, F} 0, A}, c = {I 1, I 2, ···, I n}, k s = {F 1, F 2, ···, F n}
Output: m = {m ₁ , m ₂ ,..., M _n }

STEP 1 (Step S.31 shown in FIG. 7B)
The calculation unit 15 obtains transformed data of each reference data based on the following equation.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 61 shown in FIG.

STEP 2 (Step S.32 shown in FIG. 7B)
The calculation unit 15 obtains a solution of simultaneous equations based on the following two equations, and obtains a contribution a _j (j = 1, 2,..., N).
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . ^{_{Also, I 0p, I w 1p,}} ···, I w np , ^{_{respectively, I 0, I w 1,}} ···, shows a p-th element of ^I _{w n.} In this process, the calculation unit 15 functions as a contribution calculation unit 62 shown in FIG.

STEP 3 (Step S.33 shown in FIG. 7B)
The calculation unit 15 extracts secret data using the following expression based on the calculated contribution a _j (j = 1, 2,..., N).
In this processing, the calculation unit 15 functions as the secret data reproduction unit 63 shown in FIG. Further, the secret data obtained by this process is stored in the secret data storage unit 64.

  As described above, the secret data can be extracted (restored or reproduced) from the morphing data (stego object s) through the steps 1 to 3 described above (the process shown in FIG. 7B).

  As the secret data in the second embodiment, data that cannot be directly morphed (for example, data expressed by numerical values) is used. By the above-described processing, data that cannot be morphed is converted to secret data by steganography based on morphing. As cover data. Moreover, since the secret data can be concealed using a plurality of cover data by the above-described processing, the security of the concealed secret data can be improved.

The following two formulas
The simultaneous equations consisting of can be expressed as follows, where N _e is the number of equations.
If the number of unknown secret data m is N _u , N _u = n. Usually, N _u ≦ N _e is set to uniquely determine the unknown. That, N _e will indicate the upper limit of the secret data can be concealed.

In this case, in theory,
If you choose any of N _u number of equations from, it is possible to obtain the secret data.

  Here, in practice, the morphing data may be converted into an integer value and stored, or an error may occur in the data in the process in which the stego object s is transmitted through the open network 3. For this reason, the solution of simultaneous equations may not exist strictly. In such a case, a solution can be obtained by obtaining a least square solution or using an error correction code.

Further, the stego object s, but contains the F ₀ and A, it is also possible to integrate the I ₀ it as metadata. That is, the stego object s can be output as one file from the stego object storage unit 50 of the secret information hiding device 1 or stored in the stego object storage unit 67 of the secret information restoration device 2. Furthermore, a contribution to use when determining the F _0, since it is also possible to distinguish between contribution to use when determining the morphing data, it is also possible to include seeking F ₀ in advance stego key.

Actually, a reference data set Ω including n or more reference data is prepared, and by selecting n reference data from Ω for arbitrary secret data, it can be used as reference data in the above-described processing. it can. In this case, the index (j ₁ , j ₂ ,..., J _n ) of the selected reference data and the number of secret data are also part of the stego key. Further, it is assumed that all the n pieces of secret data are numerical data, but even if they are not numerical data originally, they are represented numerically in the computer, so it is not always necessary to limit the secret data to numerical values.

The method described above is described in Document A below.
[Literature A] J. Smith and B. Comiskey, "Modulation and information hiding in images," Proc. First International Workshop on Information Hiding; Lecture Notes in Computer Science, Springer, Vol. 1174, pp. 207-227, 1996.
It is very similar to the spread spectrum steganography (hereinafter referred to as SSS method) proposed in.

The SSS method, first set of basis vectors that are orthogonal to each other _{{φ 1, φ 2, ···} , φ n} prepared, any secret data _{m = {m 1, m 2} , ···, m n} Based on the above, e is obtained by the following equation.

Further, it is assumed that the cover data c is orthogonal to all the basis vectors, and the stego object s is obtained as follows.

In order to extract (reproduce) the secret data m from the stego object s, an inner product of the stego object s and each base vector may be taken. That is
Thus, secret data m _j (j = 1, 2,..., N) is obtained.

In this SSS method, since the stego object s is a linear combination of the cover data c and the basis vector, the cover data c and the basis vector must use the blind source separation (BSS) technique shown in the following document B. May be inferred from the stego object s.
[Literature B] JF Cardoso, “Blind signal separation: statistical principles,” Proceedings of the IEEE, Special Issue on Blind Identification and Estimation, Vol. 9, No. 10, pp. 2009-2025, 1998.

  On the other hand, the method proposed in the third embodiment performs linear combination after morphing (non-linear transformation) non-orthogonal reference data. For this reason, if there is no information about the feature vector of the reference data, it is very difficult to estimate the reference data by a method such as the BSS technique. Therefore, it can be determined that the method proposed in Embodiment 3 is more secure than the SSS method.

[Embodiment 3]
Next, as Embodiment 3, when the number of secret data (number, number of data) is relatively large within the range of the number of elements (Nd) of reference data, the reference data is used by using steganography based on morphing. A process for concealing and extracting secret data will be described.

  As shown in the first embodiment and the second embodiment, if the number of reference data (number) can be sufficiently prepared to correspond to the number (number) of secret data, the secret data is concealed in the reference data. Can do. However, in the methods shown in the first and second embodiments, when the number (number) of secret data increases (when it is relatively large), the following problems occur.

1) It is necessary to prepare a large amount of reference data corresponding to a large number of secret data.
2) In steganography based on morphing, it is necessary to transform all reference data by morphing. Therefore, even if the secret data is concealed (embedded) or extracted, a huge amount of calculation is required. May be required and the calculation cost may be high.

  In order to solve these problems, a method of storing a large amount of information in a secret place and concealing only the address and pointer as secret data can be considered. Alternatively, a method of encrypting a large amount of information and concealing only the decryption key as secret data can be considered. In this way, instead of directly concealing the information to be concealed as secret data directly in the reference data, by using indirect information for obtaining the information to be concealed as small secret data, the secret data ( (Indirect information) can be concealed.

  However, it is not a method of concealing secret data in a small number of reference data using indirect information as secret data, but wants to conceal information directly in reference data using a large number (large number) of secret data. There are many needs. Furthermore, there is a great demand for reducing the number of reference data used for the concealment process while concealing information using a large number (a large number) of secret data. In the third embodiment, in order to meet such a demand, a large number (large number) of information is directly used as secret data, and the concealment of a large number of secret data is directly concealed using a small number of reference data. Also, a method of extracting will be described.

  First, assuming that the number (number) of reference data is n and the number (number) of secret data is ν, it is assumed that the number of secret data is larger than the number of reference data. Set to. The secret data is ν data obtained by digitizing different types of secret elements in the same format.

In this way, after setting the number n of reference data and the number ν of secret data,
When,
If the contribution a is variable for each “element”, it is possible to conceal a total of N _e × n secret data in the reference data. However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . ^{_{Also, I 0p, I w 1p,}} ···, I w np , ^{_{respectively, I 0, I w 1,}} ···, shows a p-th element of ^I _{w n.}

Where _Ne is as already explained
And shows the number of equations in the above two equations.

However, even if secret data having a number larger than the number N _e of the above-described equation is concealed in the reference data, it is difficult to faithfully reproduce (extract) the concealed secret data, so that v ≦ N _e To do.

Also, N _f is usually much smaller than N _d, so without losing generality,
It is sufficient to use only. Therefore, the ν ≦ _{N d.}

In order to conceal ν secret data in the reference data, it is necessary to conceal the secret data in at least ν different formulas. There are various methods for satisfying this condition. As an example, a method for concealing one secret data in one equation will be described.
By using the above, the calculation unit 15 can conceal the secret data x as follows with the secret data x.
However, J ₁ and j ₂ are natural numbers satisfying 1 ≦ j ₁ <j ₂ ≦ n, and I ⁰ _0p and I _0p are morphing data before and after concealing secret data, respectively. The p-th element is shown. _{Further, a 1, a 2, ···} , a n indicates the degree of contribution given in advance. The secret data x is assumed to be digitized.

As shown in the equation (19) of I _0p described above, first, the secret data x is concealed to I _0p by adding to the contribution a _j1 of the j _1st reference data. Next, a condition that the sum of contributions of all reference data is 1 by subtracting the secret data x from the contribution a _{j2 of} the j _2nd reference data: a ₁ , + a ₂ +... + A _n = 1
Will be satisfied.

Mentioned above
In the case where the secret data x is concealed, if the value of the secret data x is large, the value of the morphing data may be greatly changed, and the morphing data I _0p may become unnatural as a whole. Such a problem can be solved by performing concealment after scaling the secret data x to be small.

On the other hand,
As can be seen from the above, if the p th elements of I ^w _j1 and I ^w _j2 are equal, that is,
When (I ^w _{j1, p} −I ^w _{j2, p} ) = Δ _p (j ₁ , j ₂ ) = 0, even if the secret data x is concealed, the secret data concealed later is extracted (restored) Cannot be reproduced).

  Although there are a plurality of methods for solving such a problem, Embodiment 3 will be described with reference to the following example.

First, I ^w _j1 and I ^w _j2 whose p-th elements are not equal are selected to conceal secret data. If such j ₁ and j ₂ do not exist, the value of p is increased and a determination is made based on the increased p. In this way, the number of secret data x that can conceal somewhat than N _d, less.

Next, based on the above description, the calculation unit 15 conceals a plurality of secret data m = {m ₁ , m ₂ ,..., M _v } in the morphing data to generate a stego object s, A process for extracting the secret data m from the stego object s will be described.

FIG. 9 is a diagram illustrating functional blocks when the calculation unit 15 reads a program stored in the ROM 16 and performs a secret data concealment process (realizes a concealment process method) in the third embodiment. In FIG. 9, a data transformation unit (data transformation unit) 71, an F ₀ generation unit (F ₀ generation unit) 72, a substantial cover data generation unit (substantial cover data generation unit) 73, and a stego object generation unit ( The stego object generating means) 74 substantially corresponds to the calculation unit 15 that performs concealment processing (implementing concealment processing method) according to a program. Further, the secret data storage unit 75, the cover data storage unit 76, the stego key storage unit 77, the stego object storage unit 78, and the working memory 79 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Incidentally, the cover data storage unit 76, are stored in advance cover data c is the stego key storage unit 77, it is possible to assume that previously stego key k _s are stored. Also, the secret data storage unit 75 stores secret data m in advance, and the stego object storage unit 78 stores the generated stego object s.

  Further, the calculation unit 15 may acquire cover data, secret data, or a stego key used for the secret data concealment process via the open network 3 or the closed network. In this case, the calculation unit 15 functions as a cover data acquisition unit, a secret data acquisition unit, or a stego key acquisition unit, and stores the acquired data as a cover data storage unit 76, a secret data storage unit 75, or a stego key storage, respectively. The process memorize | stored in the part 77 is performed.

  However, when the secret information concealment device 1 shown in FIG. 9 is actually used, the cover data and the stego key are often stored in the cover data storage unit 76 and the stego key storage unit 77 in advance. It is done.

  Further, the calculation unit 15 functions as a stego object output unit as necessary, reads a stego object from the stego object storage unit 78, and transmits it to the computer functioning as the secret information restoring device 2 via the open network 3. It plays the role to output.

  Next, the secret data concealment process of the calculation unit 15 according to the third embodiment will be described with reference to the flowchart shown in FIG.

In the secret data concealment process according to the third embodiment, the definition is as follows.
Secret data m = {m ₁ , m ₂ ,..., M _v }, cover data c = {I ₁ , I ₂ ,..., I _n }, stego keys k _s = {F ₁ , F ₂ , .., F _n ; a ₁ , a ₂ ,..., A _n ; A; ν}, stego object s

_Here, m 1 ~m _ν is secret _data, I 1 ~I _n reference _data, F 1 to F _n is the feature vector corresponding to each of the reference data _{I 1 ~I n, a 1 ~a} n is respectively The contribution in the reference data, I ⁰ _0, is morphing data obtained from the reference data, and is the morphing data before concealing the secret data. A is a scaling factor for determining the contribution, and is a small positive positive number (for example, less than 1). Further, the relationship of ν> n is established between ν and n.
Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of outputting the stego object s.

[Algorithm for realizing function f for concealing secret data]
Input: m = {m ₁ , m ₂ ,..., M _v }, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,. F _n ; a ₁ , a ₂ ,..., A _n ; A; ν}
Output: s

STEP 1 (Step S.41 shown in FIG. 10)
First, the calculation unit 15
Is used to obtain the feature vector of the morphing data. Here, a ₁ , a ₂ ,..., _An are values normalized in advance. F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 72 shown in FIG.

STEP 2 (step S.42 shown in FIG. 10)
Next, the calculation unit 15
Is used to obtain the transformed data of each reference data. Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 71 shown in FIG.

STEP 3 (Step S.43 shown in FIG. 10)
The calculation unit 15 obtains the morphing data I ⁰ ₀ before concealing the secret data,
Ask based on. _{^{However, I 0 0p, I w 1p}} , ···, I w np , _{^{respectively, I 0 0, I w 1}} , ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as a substantial cover data generation unit 73 shown in FIG.

STEP 4 (Step S.44 shown in FIG. 10)
The calculation unit 15 scales the secret data by the following formula.
However, the scaling factor A is a positive real number that is small (for example, less than 1) as described above.

STEP 5 (step S.45 shown in FIG. 10)
Next, the calculation unit 15 sets values of p = 1, q = 1, and s = I ⁰ ₀ , and initializes the data.

STEP 6 (Yes in step S.46 shown in FIG. 10, step S.47)
The calculation unit 15 determines whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist (FIG. Step S.46 shown in FIG.

When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (Yes in step S.46), the calculation unit 15 uses j ₁ and j ₂ according to a predetermined rule. select the door, to update the stego-object _{s p} in the following equation (21) (step S.47), the process proceeds to later processing STEP8 (step S.49).

STEP 7 (No in step S.46 shown in FIG. 10, step S.48)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.46), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.48), the process proceeds to STEP 6 (step S.46 shown in FIG. 10).

STEP 8 (step S.49 shown in FIG. 10)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.49). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.50). If q <ν is satisfied (Yes in step S.50), the processing is performed in STEP6 (FIG. 10). The process proceeds to step S.46). On the other hand, when q <ν is not satisfied (in the case of No in step S.50), the calculation unit 15 finally ends the process.

In the processing of STEP4 to STEP8 described above, the calculation unit 15 functions as the stego object generation unit 74 shown in FIG. By executing the above-described processing of STEP1 to STEP8, even if the number of secret data ν is relatively large within the range of the number Nd of reference data, the morphing data is obtained by using steganography based on morphing. It becomes possible to conceal the secret data m in I ⁰ ₀ . In addition, if different types of secret elements are expressed by numerical values in the same format (for example, unsigned 8-bit integers), it is possible to conceal the secret data from the reference data using steganography based on morphing. Become.

In STEP 6, various methods are conceivable as methods for selecting j ₁ and j _2. For example, in the case where j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 are searched in order. to, a method of utilizing the first j ₁ and j ₂ found is considered a method in which utilizing a j ₁ and j ₂ as obtained morphing data is closest to an integer after hiding secret data .

Moreover, as can be seen from equation (21), stego object _{s p} is the morphing data to (substantially cover ^data) _{I 0 0p} and private data ^{_{x q (I w j1, p}} -I w j2, p) Is obtained by adding the products of Therefore, by substituting (I ^w _{j1, p} −I ^w _{j2, p} ) with Δ _p ,
^{_{s p = I 0 0p + x}} q Δ p (21 ')
Can be shown. In this case, _Δp is not limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and is generally defined as a function of reference data, reference data deformation, and substantial cover data I ⁰ _0. be able to. For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example.

The calculation unit 15 can also employ a method of generating a certain pseudo-random number and selecting j ₁ and j ₂ according to the order of the pseudo-random number, or a method of obtaining _Δp . In this case, the seed of the process for generating the pseudo-random number is a part of the stego object or the stego key.

Further, Δ _p (j ₁ , j ₂ ) ≠ 0, which is the condition shown in STEP 6 and STEP 7 (in step S.46 in FIG. 10), Δ _p (j ₁ , j ₂ ) = T or Δ It is also possible to change to the condition _p (j ₁ , j ₂ ) εT (or Δ _p ≠ 0 to Δ _p = T or Δ _p εT). Here, T is a designated value or a range of designated values, and becomes a part of a stego object or a stego key. In this way, it is possible to hide the secret data only at the designated location. In addition, it is possible to use a location where the secret data is not concealed to detect falsification or error. In addition, different (kind) secret data can be hidden in different places on the same stego object.

Further, when the number n of reference data is ₂ , selection of j ₁ and j ₂ becomes unnecessary. In this case, the number of data that can be concealed does not decrease, but if two or more reference data are used, it becomes more difficult to estimate the reference data from the morphing data, and security can be further increased.

On the other hand, the secret information restoration device 2 extracts (restores and reproduces) a plurality of secret data {m ₁ , m ₂ ,..., M _v } from the stego object s generated by the concealment process described above. Is possible.

FIG. 11 is a diagram showing functional blocks when the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing an extraction processing method). In FIG. 11, a data transformation unit (data transformation unit) 81, an F ₀ generation unit (F ₀ generation unit) 82, a substantial cover data generation unit (substantial cover data generation unit) 83, and a secret data reproduction unit ( The secret data reproducing means 84 substantially corresponds to the calculation unit 15 that performs extraction processing (implementing the extraction processing method) according to a program. The cover data storage unit 85, the stego key storage unit 86, the stego object storage unit 87, the secret data storage unit 88, and the working memory 89 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

As shown in FIG. 11, the stego object storage unit 87 stores the stego object s generated by the secret information concealment device 1 shown in FIG. 9, and the cover data storage unit 85 stores the cover data c. , the stego key storage unit 86, the stored stego key k _s, the secret data storage unit 88, the secret data m is stored.

  In Embodiment 3, the calculation unit 15 functions as a stego object acquisition unit, and the stego object s generated by the process shown in FIG. 10 from the stego object storage unit 78 of the secret information concealment device 1 shown in FIG. Is obtained via the open network 3 and stored in the stego object storage unit 87.

  Further, the calculation unit 15 functions as a cover data acquisition unit and a stego key acquisition unit, and acquires the cover data and the stego key used for the secret data concealment process in the secret information concealment device 1 via the closed network. The cover data storage unit 85 and the stego key storage unit 86 are stored. A configuration in which cover data and stego keys are stored in advance in the cover data storage unit 85 and the stego key storage unit 86 of all the secret information restoring devices 2 may be adopted.

  In addition, the calculation unit 15 functions as a secret data output unit as needed, and reads secret data from the secret data storage unit 88 and displays the extracted data on the display unit 7 or the like.

Next, the secret data extraction process of the calculation unit 15 in the third embodiment will be described with reference to the flowchart shown in FIG. When extracting a plurality of secret data hidden in the reference data, the stego object, cover data, stego key, and secret data are defined as follows.
Stego object s, cover data c = {I ₁ , I ₂ ,..., I _n }, stego key k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a ₂ ,. ., A _n ; A; ν}, secret data m = {m ₁ , m ₂ ,..., M _ν }
There was thus defined, s, c, as input data to k _s, computing unit 15 performs a process of extracting secret data m.

[Algorithm for realizing function f- ¹ for extracting secret data]
Input: s, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a _{2 ,.} ; A; ν}
Output: m = {m ₁ , m ₂ ,..., M _v }

STEP 1 (step S.61 shown in FIG. 12)
The calculation unit 15 obtains the feature vector F ₀ of the morphing data I ₀ using the following equation. However, it is assumed that the contributions a ₁ , a ₂ ,... _An are normalized in advance.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 82 shown in FIG.

STEP 2 (Step S.62 shown in FIG. 12)
The calculation part 15 calculates | requires the deformation | transformation of each reference data using following Formula.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 81 shown in FIG.

STEP 3 (Step S.63 shown in FIG. 12)
The calculation unit 15 obtains morphing data I ⁰ ₀ before concealing secret data using the following equation.
_{^{However, I 0 0p, I w 1p}} , ···, I w np , _{^{respectively, I 0 0, I w 1}} , ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as a substantial cover data generation unit 83 shown in FIG.

STEP 4 (Step S.64 shown in FIG. 12)
Next, the calculation unit 15 sets values p = 1 and q = 1 and initializes the data.

STEP 5 (Yes in step S.65 shown in FIG. 12, step S.66)
It is determined whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (step S.65 shown in FIG. 12). When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 are present (Yes in step S.65), the calculation unit 15 determines the corresponding j based on a predetermined rule. select ₁ and a _{j 2,} scaled from seeking a secret data the following equation (22) (step S.66), the process proceeds to be described next STEP7 (step S.68).
STEP 6 (No in step S.65 shown in FIG. 12, step S.67)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.65), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.67), and the process proceeds to STEP 5 (Step S.65 shown in FIG. 12).

STEP 7 (Steps S.68 and S.69 shown in FIG. 12)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.68). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.69 shown in FIG. 12), and when q <ν is satisfied (Yes in step S.69 shown in FIG. 12). Then, the process proceeds to STEP 5 (step S.65 shown in FIG. 12). On the other hand, when q <ν is not satisfied (in the case of No in step S.69 shown in FIG. 12), the process proceeds to STEP 8 (step S.70 shown in FIG. 12) described below.

STEP 8 (step S.70 shown in FIG. 12)
The calculation unit 15 extracts secret data by the following equation (23) and ends the process.

  In the processing of STEP4 to STEP8, the calculation unit 15 functions as the secret data reproduction unit 84 shown in FIG. By performing such processing, a plurality of secret data can be extracted. By executing the processing of STEP 1 to STEP 8 described above, the secret data is extracted from the stego object in which ν secret data larger than the number n of the reference data is concealed using steganography based on morphing ( Restoration, reproduction). Also, secret data is extracted (restored and reproduced) using morphing-based steganography from morphing data that conceals different types of secret elements represented by numerical values in the same format (for example, unsigned 8-bit integers). ).

Furthermore, as already described in the secret data concealment process of the third embodiment, Δ _p is not limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and generally, reference data and reference data And can be defined as a function of the substantial cover data I ⁰ ₀ . For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example.

  In the case of the secret data extraction process shown in FIG. 12, there are cases where the secret data cannot be extracted (restored or reproduced) strictly in consideration of calculation errors, transfer errors, and the like. However, when the secret data is an image or music, even if there is a slight error, the error can be allowed because the visual or auditory influence of the error is small. In addition, strictly seeking extraction (restoration and reproduction) of secret data can be handled by employing an error correction code.

  Further, a reference data set Ω including n or more reference data is prepared, and n reference data can be selected from Ω and used for arbitrary secret data. In this case, the index of the selected reference data and the number of reference data can be part of the stego key.

  In the description so far, it has been assumed that there is enough space for storing secret data. In an actual application, the storage location that satisfies the setting condition may be smaller than ν. In this case, the secret data may be divided and the concealment (restoration) process may be performed for each part.

In the third embodiment, the case has been described where the number of secret data is relatively large within the range of the number Nd of reference data elements. If the number of the secret data relative to the number of the reference data is large, as shown in equation (21) the secret data x _q, and concealed morphing data.

However, the method of concealing secret data using equation (21) can be used even when the secret data is information that cannot be morphed. If general information that cannot be morphed can be concealed, the practicality can be further enhanced. Actually, the secret data can be concealed in the morphing data using the equation (21) as the data x _q representing an arbitrary numerical value (if the secret data is not a numerical value, it is converted into a numerical value). In Equation (21), the substantially morphed data is I ⁰ _0p , and the secret data x _q is concealed in the morphing data I ⁰ _0p without performing morphing. For this reason, even if the arbitrary secret data x is information that cannot be morphed, it is possible to conceal the secret data x from the morphing data I ⁰ ₀ and generate the stego object s by digitizing the secret data x. It becomes.

In the third embodiment, when the large value of the secret data x _q is compared to the value of the morphing data, since there is a risk that significant deviations of the value of the stego object, previously secret data x _q An explanation was given of the case of scaling to a smaller value. However, when the secret data has a very small value, for example, when the secret data has a value smaller than 1, it may not feel unnatural even if the secret data is concealed in the morphing data. Therefore, when the value of the secret data x _q is small, the equation (20)
Or
It is possible to conceal the secret data value x _q as it is in the morphing data without performing the scaling process as described above. In this case, the scaling factor A or the like is not necessary.

[Embodiment 4]
Formula (21) shown in the third embodiment
Is obtained by adding the product of the secret data x _q and (I ^w _{j1, p} −I ^w _{j2, p} ) to the morphing data (substantial cover data) I ⁰ _{0p in} which the secret data is not concealed. and a feature that the object s _p is generated.

Secret data _{x q m} q, expressed as ^{_{(I w j1, p -I w}} j2, p) a delta _p, wherein the right-hand side of ^{_{(21) x q (I w}} j1, p -I w j2, p) Can be expressed as m _{q Δp} . In this case, equation (21) is
^{_{s p = I 0 0p + m}} q Δ p (24)
Can be shown.

As is apparent from equation (24), the product m _q delta _p increases, stego object s _p becomes unnatural. In FIG. 13, two face images shown in (a) indicate reference images (reference data), and (b) shows a morphing image generated from the two face images (reference images) shown in (a). (Morphing data, corresponding to morphing data I ⁰ ₀ in equation (21)). On the other hand, (d) is a secret image before concealment (secret data, corresponding to the secret data x in the equation (21)), and (c) is the secret image shown in (d) in (b). stego image generated by concealing the indicated morphing image shows (stego object, corresponds to the stego object s _p in equation (21)). FIG. 13E shows a secret image extracted from the stego image shown in FIG.

  Compared to the morphing image shown in FIG. 13B, the stego image (FIG. 13C) in which the secret image (FIG. 13D) is concealed has a noticeable unnaturalness. For this reason, the stego image (FIG. 13C) is likely to attract the attention of a third party due to its unnaturalness, and the possibility of becoming a target of an attack increases. That is, if the stego image (FIG. 13C) is concealed with the secret image (FIG. 13D), the possibility of being noticed increases. Of course, if the secret image (FIG. 13 (d)) is encrypted and the reference image and the like are managed well, even if the existence of the secret image is inferred from the unnatural image (data), the secret image can be easily obtained. Extraction cannot be performed. However, the fact that the existence of the secret image is known is a matter that should be prevented from the gist of the present invention for concealing information, and there is a risk of reducing the utility value.

Further, in order to reduce the product m _{q Δp} so that the stego image does not become unnatural, the secret data is stored using the scaling factor A or the like as in the equation (20) shown in the third embodiment. A method of setting a small value in advance is conceivable. An effective method is to conceal the secret data in advance and conceal the reduced secret data in a place where _Δp is a small value. However, this method has a problem in that concealment efficiency is reduced. Therefore, in the fourth embodiment, a method of concealing secret data using a method other than adding the product of secret data _mq and _Δp to morphing data I ⁰ _0p will be described.

In the third embodiment the formula (21), as shown in Equation (24), by adding the product of the secret data m _q and delta _p morphing data I ⁰ _0p, the case of performing the concealment of the secret data description did. However, in order to conceal the secret data, it is not always necessary to take the product of the secret data m _q and _Δp . Similarly to the conventionally known method, the secret data is directly put in the least significant b bits (for example, the least significant 4 bits (b = 4)) of the cover data (in the fourth embodiment, the morphing data I ⁰ _0p ) Thus, concealment can be performed.

In the fourth embodiment, it is assumed that secret data can be represented by a b-bit numerical value. Any secret data x, it is possible to embed the stego object s _p according to the following equation.
s _p = MSB (I ⁰ _0p ) + x (25)

However, MSB (I ⁰ _0p ) means a function for extracting upper Bb-bit data (B is a word length) in I ⁰ _0p . In the case of MSB (S _p ), it means that secret data x is extracted directly from the lower-order b bits. Thus, by embedding the secret data into the least significant b bits of the morphing data I ⁰ _0p, stego object S _p can be prevented that an unnatural. Since the secret data is concealed for morphing data I ⁰ _0p, morphing data I ⁰ _0p will serve as substantially cover data.

Next, based on the above description, the calculation unit 15 conceals a plurality of secret data m = {m ₁ , m ₂ ,..., M _v } in the morphing data I ⁰ ₀ , or the secret data m The process of extracting the will be described.

  FIG. 9 is a diagram illustrating functional blocks when the calculation unit 15 reads a program stored in the ROM 16 and performs a concealment process for secret data (implementing a concealment process method) in the fourth embodiment. The secret information concealment apparatus 1 according to the fourth embodiment includes the same functional blocks as the secret information concealment apparatus 1 described in the third embodiment. Therefore, FIG. 9 used in the third embodiment is used as each block configuration of the fourth embodiment, and the same reference numerals are used for the description.

9, the data transformation unit 71, and the F ₀ generation unit 72, substantially covering the data generator 73 and the stego object generation unit 74, (to achieve a concealment method) performs concealment processing according to a program calculating section 15 Is substantially applicable. Further, the secret data storage unit 75, the cover data storage unit 76, the stego key storage unit 77, the stego object storage unit 78, and the working memory 79 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Incidentally, the cover data storage unit 76, are stored in advance cover data c is the stego key storage unit 77, it is possible to assume that previously stego key k _s are stored. Also, the secret data storage unit 75 stores secret data m in advance, and the stego object storage unit 78 stores the generated stego object s.

  Further, the calculation unit 15 may acquire cover data, secret data, or a stego key used for the secret data concealment process via the open network 3 or the closed network. In this case, the calculation unit 15 functions as a cover data acquisition unit, a secret data acquisition unit, or a stego key acquisition unit, and stores the acquired data as a cover data storage unit 76, a secret data storage unit 75, or a stego key storage, respectively. The process memorize | stored in the part 77 is performed.

  However, when the secret information concealment device 1 shown in FIG. 9 is actually used, the cover data and the stego key are often stored in the cover data storage unit 76 and the stego key storage unit 77 in advance. It is done.

  Further, the calculation unit 15 functions as a stego object output unit as necessary, reads a stego object from the stego object storage unit 78, and transmits it to the computer functioning as the secret information restoring device 2 via the open network 3. It plays the role to output.

  Next, the secret data concealment process of the calculation unit 15 according to the fourth embodiment will be described with reference to the flowchart shown in FIG.

In the secret data concealment process according to the fourth embodiment, the definition is as follows.
Secret data m = {m ₁ , m ₂ ,..., M _v }, cover data c = {I ₁ , I ₂ ,..., I _n }, stego keys k _s = {F ₁ , F ₂ , .., F _n ; a ₁ , a ₂ ,..., A _n }, stego object s = {s ₁ , s _{2 ,.
Here,} m 1 ~m _ν is secret _data, I 1 ~I _n reference _data, F 1 to F _n is the feature vector corresponding to each of the reference data _{I 1 ~I n, a 1 ~a} n is respectively This is the degree of contribution in the reference data. Further, s _{1 to} s _ν are stego objects and are obtained based on morphing data I ⁰ ₀ obtained from reference data.

Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of outputting the stego object s.

[Algorithm for realizing function f for concealing secret data]
Input: m = {m ₁ , m ₂ ,..., M _v }, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,. F _n ; a ₁ , a ₂ ,..., A _n }
Output: s = {s ₁ , s ₂ ,..., S _v }

STEP 1 (step S.81 shown in FIG. 14)
First, the calculation unit 15
Is used to obtain the feature vector of the morphing data. Here, a ₁ , a ₂ ,..., _An are values normalized in advance. F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 72 shown in FIG.

STEP 2 (Step S.82 shown in FIG. 14)
Next, the calculation unit 15
Is used to obtain the transformed data of each reference data. Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 71 shown in FIG.

STEP 3 (Step S.83 shown in FIG. 14)
The calculation unit 15 obtains morphing data I ⁰ ₀ before concealing secret data.
_{^{However, I 0 0p, I w 1p}} , ···, I w np , _{^{respectively, I 0 0, I w 1}} , ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as a substantial cover data generation unit 73 shown in FIG.

STEP 4 (Step S.84 shown in FIG. 14)
Next, the calculation unit 15 sets a value of p = 1 and initializes the data.

STEP 5 (step S.85 shown in FIG. 14)
Then, the calculation unit 15
s _p = MSB (I ⁰ _0p ) + m _p (26)
Find Stego object s based on

STEP 6 (Step S.86, Step S.87 shown in FIG. 14)
Thereafter, the calculation unit 15 increases the value of p by 1 (p = p + 1: step S.86), and determines whether p is smaller than ν (p <ν) (step S.87). If p <ν is satisfied (Yes in step S.87), the process proceeds to STEP 5 (step S.85 shown in FIG. 14). On the other hand, when p <ν is not satisfied (in the case of No in step S.87), the calculation unit 15 finally ends the process.

In the processing of STEP 4 to STEP 6 described above, the calculation unit 15 functions as the stego object generation unit 74 shown in FIG. By executing the processing of STEP 1 to STEP 6 described above, it is possible to conceal the ν secret data in the morphing data I ⁰ ₀ .

On the other hand, the secret reconstruction device 2, stego generated by the above-described concealment processing object _{s = {s 1, s 2} , ···, s ν} from a plurality of secret data _{{m 1, m} 2, · .., M _v } can be extracted (restored, reproduced).

  FIG. 15 is a diagram illustrating functional blocks in the case where the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing an extraction processing method). In FIG. 15, the secret data reproduction unit 102 substantially corresponds to the calculation unit 15 that performs extraction processing (implementing the extraction processing method) according to a program. The stego object storage unit 101 and the secret data storage unit 103 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

  In the fourth embodiment, when the secret data m is extracted from the stego object s, the cover data and the stego key are not required. As already described, since the secret data in the fourth embodiment is concealed in the least significant b bits of the stego object, the secret data m is extracted directly from the stego object s without performing the reverse morphing process. Is possible. For this reason, the cover data and the stego key required for the reverse morphing process are not required for extracting the secret data, and the cover data storage unit, the stego key storage unit, the cover data acquisition unit, and the stego key acquisition No part is required.

  As shown in FIG. 15, the stego object storage unit 101 stores the stego object s generated by the secret information concealment device 1 shown in FIG. 9 by the processing method shown in FIG. 14, and the secret data storage unit 103 stores the stego object s. Secret data m is stored. In the fourth embodiment, the calculation unit 15 functions as a stego object acquisition unit, and from the stego object storage unit 78 of the secret information concealment device 1 shown in FIG. 9 via the open network 3 or via a closed network. The stego object s is acquired and stored in the stego object storage unit 101. Further, the calculation unit 15 functions as a secret data output unit as necessary, and reads secret data from the secret data storage unit 103 and displays the extracted data on the display unit 7 or the like.

Next, the secret data extraction process of the calculation unit 15 in the fourth embodiment will be described with reference to the flowchart shown in FIG. When extracting a plurality of secret data concealed in a stego object, a stego object and secret data are defined as follows.
Stego object s = {s ₁ , s ₂ ,..., S _v }, secret data m = {m ₁ , m _{2 ,.}
The calculation unit 15 uses the stego object s as input data, and performs a process of extracting secret data m from the stego object s.

[Algorithm for realizing function f- ¹ for extracting secret data]
Input: s = {s ₁ , s ₂ ,..., S _v }
Output: m = {m ₁ , m ₂ ,..., M _v }

STEP 1 (step S.91 shown in FIG. 16)
First, the calculation unit 15 sets a value of p = 1 and initializes data.

STEP 2 (step S.92 shown in FIG. 16)
Calculation unit 15, the secret data _{m p,} (Step S.92) from by the following equation (27), the process proceeds to be described next STEP3 (Step S.93).
m _p = s _p −MSB (s _p ) (27)

STEP 3 (Step S.93 shown in FIG. 16)
The calculation unit 15 increases the value of p by 1 (p = p + 1: step S.93). Then, the calculation unit 15 determines whether or not p <ν is satisfied (step S.94 shown in FIG. 16), and when p <ν is satisfied (Yes in step S.94 shown in FIG. 16). Shifts the processing to STEP2 (step S.92 shown in FIG. 16). On the other hand, if p <ν is not satisfied (No in step S.94 shown in FIG. 16), the process is terminated. By performing such processing, a plurality of secret data can be extracted from the stego object.

  FIG. 17A shows a stego image (stego object) in which the secret image (secret data) shown in FIG. 17B is concealed in the least significant b bits as described above, and FIG. The secret images extracted from (a) are shown. When the stego image shown in FIG. 17A and the stego image shown in FIG. 13C are compared, the same secret image (the secret image shown in FIG. 13D and FIG. 17B) is concealed. Nevertheless, the stego image in FIG. 17A is a natural face image compared to the stego image in FIG. 13C, and is unnatural as shown in FIG. There is not. For this reason, by using the stego image shown in FIG. 17A, it becomes difficult for a third party to notice the concealment of the secret data, and it is possible to reduce the possibility of being the target of the attack.

  In addition, the extracted secret image shown in FIG. 17C is less likely to cause data deterioration due to concealment and extraction, as is clear as compared with the secret image shown in FIG. It becomes possible.

[Embodiment 5]
Next, a fifth embodiment will be described. Embodiment 4 described above is characterized in that the secret data is concealed in the least significant b bits of the morphing data, so that cover data and a stego key are not required to extract the secret data. For this reason, if there is a stego object, anyone can extract secret data, and even if the data has been tampered with, the side that acquired the stego object cannot confirm tampering. there were.

  In the fifth embodiment, even when the secret data is concealed in the least significant b bits of the morphing data, a case will be described in which cover data and a stego key are required when extracting (restoring) the secret data. As with the first to third embodiments, when the secret data is concealed and extracted (restored), the cover data and the stego key are required to prevent the secret data from being easily extracted, When secret data is extracted, it becomes possible to check whether the data has been tampered with.

Specifically, the fifth embodiment is characterized in that the location where the secret data is embedded (hidden) is controlled based on the value of _Δp . That is, only when the value of _Δp takes a predetermined value,
s _p = MSB (I ⁰ _0p ) + x (25)
The secret data x is embedded (concealed) in the least significant b bits of I ⁰ _0p . When the value of _Δp does not take a predetermined value, the value of the morphing data (substantial cover data) I ⁰ _0p can be used as it is, or predetermined “confirmation information” can be embedded.

Next, based on the above description, the calculation unit 15 conceals or extracts a plurality of secret data m = {m ₁ , m ₂ ,..., _{M v} } from the morphing data I ⁰ _0p. Give an explanation.

  FIG. 9 is a diagram illustrating functional blocks in the case where the calculation unit 15 reads a program stored in the ROM 16 and performs a concealment process of secret data (realizes a concealment process method) in the fifth embodiment. The secret information concealment apparatus 1 according to the fifth embodiment includes the same functional blocks as the secret information concealment apparatus 1 described in the third embodiment. Therefore, FIG. 9 used in the third embodiment is used as each block configuration of the fifth embodiment, and the same reference numerals are used for the description.

In FIG. 9, a data transformation unit (data transformation unit) 71, an F ₀ generation unit (F ₀ generation unit) 72, a substantial cover data generation unit (substantial cover data generation unit) 73, and a stego object generation unit ( The stego object generating means) 74 substantially corresponds to the calculation unit 15 that performs concealment processing (implementing concealment processing method) according to a program. Further, the secret data storage unit 75, the cover data storage unit 76, the stego key storage unit 77, the stego object storage unit 78, and the working memory 79 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Incidentally, the cover data storage unit 76, are stored in advance cover data c is the stego key storage unit 77, it is possible to assume that previously stego key k _s are stored. Also, the secret data storage unit 75 stores secret data m in advance, and the stego object storage unit 78 stores the generated stego object s.

  Further, the calculation unit 15 may acquire cover data, secret data, or a stego key used for the secret data concealment process via the open network 3 or the closed network. In this case, the calculation unit 15 functions as a cover data acquisition unit, a secret data acquisition unit, or a stego key acquisition unit, and stores the acquired data as a cover data storage unit 76, a secret data storage unit 75, or a stego key storage, respectively. The process memorize | stored in the part 77 is performed.

  However, when the secret information concealment device 1 shown in FIG. 9 is actually used, the cover data and the stego key are often stored in the cover data storage unit 76 and the stego key storage unit 77 in advance. It is done.

  Further, the calculation unit 15 functions as a stego object output unit as necessary, reads a stego object from the stego object storage unit 78, and transmits it to the computer functioning as the secret information restoring device 2 via the open network 3. It plays the role to output.

  Next, the secret data concealment process of the calculation unit 15 according to the fifth embodiment will be described with reference to the flowchart shown in FIG.

In the secret data concealment process according to the fifth embodiment, the definition is as follows.
Secret data m = {m ₁ , m ₂ ,..., M _v }, cover data c = {I ₁ , I ₂ ,..., I _n }, stego keys k _s = {F ₁ , F ₂ , , F _n ; a ₁ , a ₂ ,..., A _n }, stego object s
_Here, m 1 ~m _ν is secret _data, I 1 ~I _n reference _data, F 1 to F _n is the feature vector corresponding to each of the reference data _{I 1 ~I n, a 1 ~a} n is respectively This is the degree of contribution in the reference data. Further, s is a stego object and is obtained based on morphing data I ₀ obtained from reference data.

Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of outputting the stego object s.

[Algorithm for realizing function f for concealing secret data]
Input: m = {m ₁ , m ₂ ,..., M _v }, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,. F _n ; a ₁ , a ₂ ,..., A _n }
Output: s

STEP 1 (step S.111 shown in FIG. 18)
First, the calculation unit 15
Is used to obtain the feature vector of the morphing data. Here, a ₁ , a ₂ ,..., _An are values normalized in advance. F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 72 shown in FIG.

STEP 2 (step S.112 shown in FIG. 18)
Next, the calculation unit 15
Is used to obtain the transformed data of each reference data. Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 71 shown in FIG.

STEP 3 (Step S.113 shown in FIG. 18)
The calculation unit 15 obtains the morphing data I ⁰ ₀ before concealing the secret data,
Ask based on. _{^{However, I 0 0p, I w 1p}} , ···, I w np , _{^{respectively, I 0 0, I w 1}} , ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as a substantial cover data generation unit 73 shown in FIG.

STEP 4 (step S.114 shown in FIG. 18)
Next, the calculation unit 15 sets values of p = 1, q = 1, and s = I ⁰ ₀ , and initializes the data.

STEP 5 (Yes in step S.115 shown in FIG. 18, step S.116)
The calculation unit 15 determines whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist (FIG. Step S.115 shown in FIG.
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (Yes in step S.115), the calculation unit 15 uses j ₁ and j ₂ according to a predetermined rule. select the door, the stego-object ^{_{s s p = MSB (I 0}} 0p) + m q (26)
(Step S.116), and the process proceeds to STEP 7 (Step S.118) described later. Here, s _p indicates the p-th element of s, m _q represents the q-th element of m. However, q is q ≦ ν.

STEP 6 (No in step S.115 shown in FIG. 18, step S.117)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.115), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.117), and the process proceeds to STEP 5 (Step S.115 shown in FIG. 18).

STEP 7 (Steps S.118 and S.119 shown in FIG. 18)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.118). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.119), and when q <ν is satisfied (Yes in step S.119), the processing is performed in STEP5 (FIG. 18). The process proceeds to step S.115). On the other hand, when q <ν is not satisfied (in the case of No in step S.119), the calculation unit 15 finally ends the process.

In the processing of STEP 4 to STEP 7 described above, the calculation unit 15 functions as the stego object generation unit 74 shown in FIG. By executing the processing of STEP 1 to STEP 7 described above, it becomes possible to conceal the secret data in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. It is difficult for a third party to easily determine at least where the secret data is concealed. In this case, the morphing data I ⁰ ₀ corresponds to the substantial cover data.

In the secret data concealment process of the fifth embodiment described above, Δ _p is not limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and generally, reference data and reference data are modified. Can be defined as a function of the substantial cover data I ⁰ ₀ . For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example. For example, Δ _p = (I _{j1, p} −I _{j2, p} ) can be set as Δ _p using a function of reference data. The method of calculating _Δp can be shared between the above-described secret information concealment device 1 (the information concealment side) and the secret information restoration device 2 (the information restoration side) described below, or It can also be part of the stego key.

The secret information restoration device 2 extracts (restores and reproduces) a plurality of secret data {m ₁ , m ₂ ,..., M _v } from the stego object s whose secret data is concealed by the above-described concealment process. Is possible.

FIG. 19 is a diagram illustrating functional blocks when the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing an extraction processing method). In FIG. 19, a data transformation unit (data transformation unit) 91, an F ₀ generation unit (F ₀ generation unit) 92, and a secret data reproduction unit (secret data reproduction unit) 94 perform extraction processing according to a program (extraction). The calculation unit 15 (which realizes the processing method) substantially corresponds. The cover data storage unit 95, the stego key storage unit 96, the stego object storage unit 97, the secret data storage unit 98, and the working memory 99 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

As shown in FIG. 19, the stego object storage unit 97 stores the stego object s generated by the processing method shown in FIG. 18 in the secret information concealment apparatus 1 shown in FIG. , cover the data c is stored, the stego key storage unit 96, the stored stego key k _s, the secret data storage unit 98, the secret data m is stored.

  In the fifth embodiment, the calculation unit 15 functions as a stego object acquisition unit, and the stego object s generated by the process shown in FIG. 18 from the stego object storage unit 78 of the secret information concealment device 1 shown in FIG. Is obtained via the open network 3 and stored in the stego object storage unit 97.

  Further, the calculation unit 15 functions as a cover data acquisition unit and a stego key acquisition unit, and acquires the cover data and the stego key used for the secret data concealment process in the secret information concealment device 1 via the closed network. The cover data storage unit 95 and the stego key storage unit 96 are stored. A configuration in which cover data and stego keys are stored in advance in the cover data storage unit 95 and the stego key storage unit 96 of all the secret information restoring devices 2 may be adopted.

  Further, the calculation unit 15 functions as a secret data output unit as necessary, and reads secret data from the secret data storage unit 98 and displays the extracted data on the display unit 7 or the like.

Next, the secret data extraction processing of the calculation unit 15 in the fifth embodiment will be described with reference to the flowchart shown in FIG. When extracting a plurality of secret data hidden in the reference data, the stego object, cover data, stego key, and secret data are defined as follows.
Stego object s, cover data c = {I ₁ , I ₂ ,..., I _n }, stego key k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a ₂ ,. ., A _n }, secret data m = {m ₁ , m ₂ ,..., M _v }
Defined in this manner, s, c, as input data to k _s, computing unit 15 performs a process of extracting secret data m from the stego object s.

[Algorithm for realizing function f- ¹ for extracting secret data]
Input: s, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a _{2 ,.} }
Output: m = {m ₁ , m ₂ ,..., M _v }

STEP 1 (step S.121 shown in FIG. 20)
The calculation unit 15 obtains the feature vector F ₀ of the morphing data I ₀ using the following equation. However, it is assumed that the contributions a ₁ , a ₂ ,... _An are normalized in advance.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 92 shown in FIG.

STEP2 (step S.122 shown in FIG. 20)
The calculation part 15 calculates | requires the deformation | transformation of each reference data using following Formula.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 91 shown in FIG.

STEP 3 (Step S.123 shown in FIG. 20)
Next, the calculation unit 15 sets values p = 1 and q = 1 and initializes the data.

STEP 4 (Yes in step S.124 shown in FIG. 20, step S.125)
The calculation unit 15 determines whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (step S.124 shown in FIG. 20). When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (Yes in step S.124), the calculation unit 15 determines the corresponding j based on a predetermined rule. Choose ₁ and j ₂ and scaled secret data
m _q = s _p −MSB (s _p ) (27)
(Step S.125), the process proceeds to STEP 6 (Step S.127) described later. Here, s _p indicates the p-th element of s, m _q represents the q-th element of m. However, q is q ≦ ν.

STEP 5 (No in step S.124 shown in FIG. 20, step S.126)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.124), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.126), the process proceeds to STEP 4 (step S.124 shown in FIG. 20).

STEP 6 (Steps S.127 and S.128 shown in FIG. 20)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.127). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.128 shown in FIG. 20), and when q <ν is satisfied (Yes in step S.128 shown in FIG. 20). Then, the process proceeds to STEP 4 (step S.124 shown in FIG. 20). On the other hand, if q <ν is not satisfied (No in step S.128 shown in FIG. 20), the process ends.

In the processing of STEP4 to STEP6, the calculation unit 15 functions as the secret data reproduction unit 94 shown in FIG. By performing such processing, it is possible to extract the secret data by searching for the location where the secret data is concealed from the stego object s where the plurality of secret data is concealed. In the fifth embodiment, unlike the fourth embodiment, I ^w _{j1, p} and I ^w _{j2, p} are obtained using the cover data and the stego key, and Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ p must be obtained so that even if only the stego object s is obtained, the secret data m cannot be extracted and security can be improved. It becomes.

In the secret data extraction process of the fifth embodiment described above, _Δp is not limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and generally, reference data and reference data are transformed. Can be defined as a function of the substantial cover data I ⁰ ₀ . For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example.

For example, Δ _p = (I _{j1, p} −I _{j2, p} ) can be set as Δ _p using a function of reference data. When defining as a function of the substantial cover data I ⁰ ₀ , the calculation unit 15 sets the morphing data I ⁰ ₀ before concealing the secret data,
It is possible to set _Δp by obtaining based on.

Furthermore, the method of calculating _Δp can be shared between the secret information concealment device 1 (the information concealment side) and the secret information restoration device 2 (the information restoration side) described above, or a stego key It can also be part of

[Embodiment 6]
Next, a sixth embodiment will be described. In the sixth embodiment, the method described in the fifth embodiment is further improved. Specifically, when the secret data x is embedded in the least significant bit of the morphing data, it is not embedded directly, but the bitwise XOR between the numerical value y related to the reference data, the morphing data, etc. and the secret data x It is characterized by embedding after performing (exclusive OR) operation. That is,
s _p = MSB (I ⁰ _0p ) + x (25)
Change to

Here, a character in which “+” is superimposed on + indicates a bitwise XOR operation, x is an element of secret data indicated by b bits, and y is expressed by b bits obtained from cover data, reference data or a modification thereof. Indicates possible numbers. As an example corresponding to y, the following may be used.
1) y = LSB (I ⁰ _0p )... Y is the least significant b bit of I ⁰ _0p .
^{2) y = LSB (I w} 1, p -I w 2, p)
_{^{··· I w 1, p -I w}} 2, the least significant b bits of _p are as y.
3) y = LSB (I ^w _{j, p} )... _Let y be the least significant b bits of I ^w _{j, p} .
In other words, as y, I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), which can be set to data consisting of a b-bit numerical value. In addition, the data is not necessarily limited to the least significant b bits as long as the data is composed of numerical values of b bits, but in the sixth embodiment, description is given using the data of the least significant b bits as an example.

In this way, by setting y, the third party can embed the secret data in a form depending on the reference data or morphing data, so that the third party can start from the least significant bit of the stego object s.
Even if is extracted, x cannot be obtained. Further, since y serves as an encryption key, it is not necessary to encrypt the secret data before embedding the secret data. For this reason, it is possible to omit the calculation cost of encryption.

Next, based on the above description, a description will be given of a process in which the calculation unit 15 conceals or extracts a plurality of secret data m = {m ₁ , m ₂ ,..., M _v } from the cover data c. Do.

  FIG. 9 is a diagram illustrating functional blocks when the calculation unit 15 reads a program stored in the ROM 16 and performs a secret data concealment process (realizes a concealment process method) in the sixth embodiment. The secret information concealment apparatus 1 according to the sixth embodiment includes the same functional blocks as the secret information concealment apparatus 1 described in the third embodiment. Therefore, FIG. 9 used in the third embodiment is used as each block configuration of the sixth embodiment, and the same reference numerals are used for the description.

In FIG. 9, a data transformation unit (data transformation unit) 71, an F ₀ generation unit (F ₀ generation unit) 72, a substantial cover data generation unit (substantial cover data generation unit) 73, and a stego object generation unit ( The stego object generating means) 74 substantially corresponds to the calculation unit 15 that performs concealment processing (implementing concealment processing method) according to a program. Further, the secret data storage unit 75, the cover data storage unit 76, the stego key storage unit 77, the stego object storage unit 78, and the working memory 79 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

Incidentally, the cover data storage unit 76, are stored in advance cover data c is the stego key storage unit 77, it is possible to assume that previously stego key k _s are stored. Also, the secret data storage unit 75 stores secret data m in advance, and the stego object storage unit 78 stores the generated stego object s.

  Further, the calculation unit 15 may acquire cover data, secret data, or a stego key used for the secret data concealment process via the open network 3 or the closed network. In this case, the calculation unit 15 functions as a cover data acquisition unit, a secret data acquisition unit, or a stego key acquisition unit, and stores the acquired data as a cover data storage unit 76, a secret data storage unit 75, or a stego key storage, respectively. The process memorize | stored in the part 77 is performed.

  However, when the secret information concealment device 1 shown in FIG. 9 is actually used, the cover data and the stego key are often stored in the cover data storage unit 76 and the stego key storage unit 77 in advance. It is done.

  Further, the calculation unit 15 functions as a stego object output unit as necessary, reads a stego object from the stego object storage unit 78, and transmits it to the computer functioning as the secret information restoring device 2 via the open network 3. It plays the role to output.

  Next, the secret data concealment process of the calculation unit 15 according to the sixth embodiment will be described with reference to the flowchart shown in FIG.

In the secret data concealment process according to the sixth embodiment, the definition is as follows.
Secret data m = {m ₁ , m ₂ ,..., M _v }, cover data c = {I ₁ , I ₂ ,..., I _n }, stego keys k _s = {F ₁ , F ₂ , , F _n ; a ₁ , a ₂ ,..., A _n }, stego object s
_Here, m 1 ~m _ν is secret _data, I 1 ~I _n reference _data, F 1 to F _n is the feature vector corresponding to each of the reference data _{I 1 ~I n, a 1 ~a} n is respectively This is the degree of contribution in the reference data. Further, s is a stego object, which is obtained based on morphing data I ⁰ ₀ obtained from reference data.

Defined in this way, m, c, as input data to k _s, computing unit 15 performs a process of outputting the stego object s.

[Algorithm for realizing function f for concealing secret data]
Input: m = {m ₁ , m ₂ ,..., M _v }, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,. F _n ; a ₁ , a ₂ ,..., A _n }
Output: s

STEP 1 (step S.131 shown in FIG. 21)
First, the calculation unit 15
Is used to obtain the feature vector of the morphing data. Here, a ₁ , a ₂ ,..., _An are values normalized in advance. F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 72 shown in FIG.

STEP 2 (Step S.132 shown in FIG. 21)
Next, the calculation unit 15
Is used to obtain the transformed data of each reference data. Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 71 shown in FIG.

STEP 3 (Step S.133 shown in FIG. 21)
The calculation unit 15 obtains the morphing data I ⁰ ₀ before concealing the secret data,
Ask based on. _{^{However, I 0 0p, I w 1p}} , ···, I w np , _{^{respectively, I 0 0, I w 1}} , ···, shows a p-th element of ^I _{w n.} In this processing, the calculation unit 15 functions as a substantial cover data generation unit 73 shown in FIG.

STEP 4 (step S.134 shown in FIG. 21)
Next, the calculation unit 15 sets values of p = 1, q = 1, and s = I ⁰ ₀ , and initializes the data.

STEP 5 (Yes in step S.135 shown in FIG. 21, step S.136)
The calculation unit 15 determines whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist (FIG. Step S.135 shown in 21)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (Yes in step S.135), the calculation unit 15 uses j ₁ and j ₂ according to a predetermined rule. select the door, the stego-object s _p
(Step S.136), and the process proceeds to STEP 7 (Step S.138) described later. y is a numerical value obtained based on I ⁰ _0p , I ^w _{j1, p} , I ^w _{j2, p, and the} like and can be represented by b bits. Further, s _p indicates the p-th element of s, m _q represents the q-th element of m. However, q is q ≦ ν.

STEP 6 (No in step S.135 shown in FIG. 21, step S.137)
When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.135), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.137), and the process proceeds to STEP 5 (Step S.135 shown in FIG. 21).

STEP 7 (step S.138 shown in FIG. 21)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.138). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.139). If q <ν is satisfied (Yes in step S.139), the processing is performed in STEP5 (FIG. 21). The process proceeds to step S.135). On the other hand, when q <ν is not satisfied (No in step S.139), the calculation unit 15 finally ends the process.

In the processing of STEP 4 to STEP 7 described above, the calculation unit 15 functions as the stego object generation unit 74 shown in FIG. By executing the processing of STEP 1 to STEP 7 described above, it becomes possible to conceal the secret data in the least significant b bits of the morphing data I ⁰ ₀ at _p where Δ _p (j ₁ , j ₂ ) ≠ 0. . Further, the secret data to be concealed is concealed after an XOR operation is performed on the b-bit value obtained from the morphing data and the like and the secret data. For this reason, it is possible to make it difficult for a third party to determine that the secret data is concealed, and it is possible to reduce the risk of the secret data being extracted by the third party.

Note that in the secret data concealment process of the sixth embodiment described above, Δ _p is not necessarily limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and in general, the reference data and the modification of the reference data Can be defined as a function of the substantial cover data I ⁰ ₀ . For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example. For example, Δ _p = (I _{j1, p} −I _{j2, p} ) can be set as Δ _p using a function of reference data. The method of calculating _Δp can be shared between the above-described secret information concealment device 1 (the information concealment side) and the secret information restoration device 2 (the information restoration side) described below, or It can also be part of the stego key.

The secret information restoration device 2 extracts (restores and reproduces) a plurality of secret data {m ₁ , m ₂ ,..., M _v } from the stego object s whose secret data is concealed by the above-described concealment process. Is possible.

  FIG. 19 is a diagram illustrating functional blocks when the calculation unit 15 reads out a program stored in the ROM 16 and performs extraction processing of secret data (implementing an extraction processing method). The secret information restoring apparatus 2 according to the sixth embodiment includes the same functional blocks as the secret information restoring apparatus 2 described in the fifth embodiment. Therefore, FIG. 19 used in the fifth embodiment is used as each block configuration of the sixth embodiment, and the same reference numerals are used for the description.

In FIG. 19, a data transformation unit (data transformation unit) 91, an F ₀ generation unit (F ₀ generation unit) 92, and a secret data reproduction unit (secret data reproduction unit) 94 perform extraction processing according to a program (extraction). The calculation unit 15 (which realizes the processing method) substantially corresponds. The cover data storage unit 95, the stego key storage unit 96, the stego object storage unit 97, the secret data storage unit 98, and the working memory 99 substantially correspond to the RAM 17 or the auxiliary storage unit 12.

As shown in FIG. 19, the stego object storage unit 97 stores the stego object s generated by the processing method shown in FIG. 21 in the secret information concealment device 1 shown in FIG. , cover the data c is stored, the stego key storage unit 96, the stored stego key k _s, the secret data storage unit 98, the secret data m is stored.

  In Embodiment 6, the calculation unit 15 functions as a stego object acquisition unit, and the stego object generated by the processing method shown in FIG. 21 from the stego object storage unit 78 of the secret information concealment device 1 shown in FIG. s is acquired via the open network 3 and stored in the stego object storage unit 97.

  Further, the calculation unit 15 functions as a cover data acquisition unit and a stego key acquisition unit, and acquires the cover data and the stego key used for the secret data concealment process in the secret information concealment device 1 via the closed network. The cover data storage unit 95 and the stego key storage unit 96 are stored. A configuration in which cover data and stego keys are stored in advance in the cover data storage unit 95 and the stego key storage unit 96 of all the secret information restoring devices 2 may be adopted.

  Further, the calculation unit 15 functions as a secret data output unit as necessary, and reads secret data from the secret data storage unit 98 and displays the extracted data on the display unit 7 or the like.

Next, the secret data extraction process of the calculation unit 15 in the sixth embodiment will be described with reference to the flowchart shown in FIG. When extracting a plurality of secret data hidden in the reference data, the stego object, cover data, stego key, and secret data are defined as follows.
Stego object s, cover data c = {I ₁ , I ₂ ,..., I _n }, stego key k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a ₂ ,. ., A _n }, secret data m = {m ₁ , m ₂ ,..., M _v }
There was thus defined, s, c, as input data to k _s, computing unit 15 performs a process of extracting secret data m.

[Algorithm for realizing function f- ¹ for extracting secret data]
Input: s, c = {I ₁ , I ₂ ,..., I _n }, k _s = {F ₁ , F ₂ ,..., F _n ; a ₁ , a _{2 ,.} }
Output: m = {m ₁ , m ₂ ,..., M _v }

STEP 1 (step S.141 shown in FIG. 22)
The calculation unit 15 obtains the feature vector F ₀ of the morphing data I ₀ using the following equation. However, it is assumed that the contributions a ₁ , a ₂ ,... _An are normalized in advance.
However, F _jp (j = 0, 1,..., N) indicates the p-th element of F _j . In this process, the calculation unit 15 functions as the F ₀ generation unit 92 shown in FIG.

STEP2 (Step S.142 shown in FIG. 22)
The calculation part 15 calculates | requires the deformation | transformation of each reference data using following Formula.
Here, I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , and W indicates a deformation (warping) function. I _j (j = 1, 2,..., N) is transformed by the deformation function W to generate I ^w _j (j = 1, 2,..., N). I ^w _j (j = 1, 2,..., N) has a feature vector F ₀ . In this process, the calculation unit 15 functions as the data transformation unit 91 shown in FIG.

STEP 3 (Step S.143 shown in FIG. 22)
Next, the calculation unit 15 sets values p = 1 and q = 1 and initializes the data.

STEP 4 (Yes in step S.144 shown in FIG. 22, step S.145)
The calculation unit 15 determines whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 exist (step S.144 shown in FIG. 22). When j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 are present (Yes in step S.144), the calculation unit 15 determines the corresponding j based on a predetermined rule. Choose ₁ and j ₂ and scaled secret data
(Step S.145), the process proceeds to STEP 6 (Step S.147) described later. However, y is a numerical value that can be obtained based on I ⁰ _0p , I ^w _{j1, p} , I ^w _{j2, p, and the} like and can be represented by b bits. Further, s _p indicates the p-th element of s, m _q represents the q-th element of m. However, q is q ≦ ν.

STEP 5 (No in step S.144 shown in FIG. 22, step S.146)
If j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) ≠ 0 do not exist (No in step S.144), the calculation unit 15 increases the value of p by 1 (p = p + 1: Step S.146), and the process proceeds to STEP 4 (Step S.144 shown in FIG. 22).

STEP 6 (Steps S.147 and S.148 shown in FIG. 22)
The calculation unit 15 increases the value of p by 1 (p = p + 1), and further increases the value of q by 1 (q = q + 1) (step S.147). Then, the calculation unit 15 determines whether or not q <ν is satisfied (step S.148 shown in FIG. 22), and when q <ν is satisfied (Yes in step S.148 shown in FIG. 22). Then, the process proceeds to STEP 4 (step S.144 shown in FIG. 22). On the other hand, if q <ν is not satisfied (No in step S.148 shown in FIG. 22), the process ends.

  In the processing of STEP4 to STEP6, the calculation unit 15 functions as the secret data reproduction unit 94 shown in FIG. By performing such processing, the secret data is concealed from the stego object s in which a plurality of secret data is concealed, the data is extracted, and the extracted data and the value of y are extracted. Since the secret data is extracted by performing the XOR operation, the security can be improved. Further, as in the fifth embodiment, it is possible to ensure sufficient security basically without performing encryption. Furthermore, if there is no communication error, the secret data can be accurately reproduced. It becomes possible to increase the extraction accuracy.

In the sixth embodiment described above, the case where a numerical value related to reference data, morphing data, or the like is used as the value of y has been described. However, y is not necessarily limited to a numerical value related to reference data, morphing data, or the like. For example, it is possible to set y by setting a seed value r and generating a random number sequence y ₁ , y ₂ ,... Based on the seed value r. In this case, it is necessary to add the seed value r to the stego key.

Also in the secret data extraction process of the sixth embodiment described above, Δ _p is not limited to (I ^w _{j1, p} −I ^w _{j2, p} ), and generally, reference data and reference data are transformed. Can be defined as a function of the substantial cover data I ⁰ ₀ . For this reason, Δ _p = (I ^w _{j1, p} −I ^w _{j2, p} ) is merely an example.

For example, Δ _p = (I _{j1, p} −I _{j2, p} ) can be set as Δ _p using a function of reference data. When defining as a function of the substantial cover data I ⁰ ₀ , the calculation unit 15 sets the morphing data I ⁰ ₀ before concealing the secret data,
It is possible to set _Δp by obtaining based on.

Furthermore, the method of calculating _Δp can be shared between the secret information concealment device 1 (the information concealment side) and the secret information restoration device 2 (the information restoration side) described above, or a stego key It can also be part of

  Next, security will be described. In the conventional method, the secret data m can be extracted from the stego object s as long as the extraction side can obtain the stego key. If the existence of the secret data is known, or if it is known that the secret data is originally hidden in the stego object s, the stego object s can be estimated by guessing a small stego key. There is a possibility that extraction of the secret data m is executed.

  However, when the secret data concealment method shown in the third embodiment, the fifth embodiment, the sixth embodiment, or the like is used, both the reference data and the stego key are required to extract the secret data m. It becomes. When combined with the reference data and the stego key, it can be regarded as a very large encryption key, and it is theoretically impossible to guess this encryption key. Accordingly, when the concealment method such as the third embodiment is used, it is very difficult to guess the secret data m from the stego object s. Of course, since the secret data can be concealed after being encrypted, it can be said that the security of the secret data can be further enhanced by performing the concealment process after the encryption.

  As described in the first to sixth embodiments, the cover data and the stego key are stored in the cover data storage unit and the stego key storage unit of the secret information concealment device 1 and the secret information restoration device 2, respectively. To share data. Then, the secret information concealment device 1 generates a stego object in which the secret data is concealed in the cover data, and transmits the stego object from the secret information concealment device 1 to the secret information restoration device 2 via the open network 3. . By adopting such a configuration, the secret information restoring apparatus 2 can extract (reproduce) secret data from a stego object based on shared cover data and a stego key.

  In addition, by transmitting / receiving a stego object in which secret data is concealed via the open network 3, it is possible to expand and use the secret data for group communication.

  For example, each member of the group can not only individually send and receive Stego objects with hidden secret data, but also post stego objects with hidden secret data on a Web bulletin board or blog. You can also In this case, it is possible for other members to extract (reproduce) secret data by downloading a stego object from the bulletin board or the like as necessary. This bulletin board or the like may be shared by a group or may be individually owned by each member.

  It is also possible to send and receive e-mails with attached stego objects between members, so that the recipient confirms the sender of the e-mail instead of passively receiving stego objects. In addition, since attached data can be acquired, spam mails and malicious mails can be avoided.

  Furthermore, by applying the secret information concealment device 1 and the secret information restoration device 2 shown in the first to sixth embodiments, a storage medium storing user (user) authentication information, such as an authentication card, etc. It is also possible to construct a card user authentication system using the.

  FIG. 23 is a diagram showing a schematic configuration of a card user authentication system. In the authentication system 110 shown in FIG. 23, at least one remote server 111 and one or more local terminals 112 are connected via an open network 3. Further, the local terminal 112 is provided with a card reader 112 a for reading data (for example, a stego key and a stego object) used for authentication of a card owned by the user from the card 113.

  In such an authentication system 110, the cover data is stored in the local terminal 112 and the remote server 111 so that the cover data is shared, and the stego key is distributed and stored in either the remote server 111 or the card 113. Let Further, the remote server 111 selects reference data as necessary for an arbitrary user, and stegoes user authentication information (for example, this information such as a face photograph and vein pattern corresponds to secret data). It is hidden (embedded) in the object, and the stego object is distributed and stored in either the remote server 111 or the card 113.

  When the user presents the card 113 to an operator who operates the local terminal 112 and the card reader 112a can read the data stored in the card 113, the local terminal 112 directly or directly from the card 113 The stego key and stego object are acquired from the remote server 111 connected to the network 3 based on the card information, and the user authentication information is restored using the cover data stored in the local terminal 112 in advance. The user authentication information restored in this way, for example, a photograph of the authentic user, is displayed on the display screen of the local terminal 112 or the like. The operator compares the displayed photograph of the authentic user's face with the face of the person who actually presented the card 113 to determine whether or not the card presenter is the owner of the authentic (proper) card 113. Can be determined.

  In the secret information concealment method and the secret information restoration method shown in the second to sixth embodiments, the stego object has a photograph from a different angle of the same person, a photograph of a relative or a substitute person, or other than the photograph. Authentication information, morphing data obtained by the method of the first embodiment, and the like can be embedded.

  FIG. 24 shows four small images shown in FIG. 24B using the two cover images shown in FIG. 24A using the secret information concealment method described in the third embodiment. The case where (secret image) is embedded in the morphing image shown in FIG. 24C is shown as an example. FIG. 24D shows an image (extracted image) extracted from the morphing image shown in FIG. Since the case where the morphing image is converted into an 8-bit integer and stored is shown, the extracted image in FIG. 24D is compared with the original image (secret image) shown in FIG. However, it is possible to determine that the deterioration is within a permissible range if the images are based on the same image.

  In the methods shown in the fourth to sixth embodiments, only the cover data is stored in the cover data storage units of the secret information concealment device 1 and the secret information restoration device 2, and the reference data is shared. It is also possible to adopt a method in which the key is hidden directly on the stego object.

  For example, in the authentication system 110 shown in FIG. 23, a plurality of reference data (cover data) can be stored in the local terminal 112 in advance as a reference data set, or the local terminal can be connected to the remote server 111 via a closed network. As described above, the reference data set stored in the remote terminal 111 can be read at any time, and the stego object in which the secret data and the stego key are concealed can be transmitted to the local terminal 112 via the open network 3. In such an authentication system 110, the secret data can be concealed and restored on the transmitting side and the receiving side of the stego object by the following procedure.

[Concealment of secret data and stego key in stego object]
Step 1: Let N be the number of members of the reference data set. n (1 <n <N) different integer random numbers i ₁ , i ₂ ,..., _n are generated, and n pieces of reference data are obtained from the reference data set using them as indexes.
Step 2: Define feature vectors F ₁ , F ₂ ,..., F _n for each reference data.
Step 3: Generate _n positive (> 0) random numbers r ₁ , r ₂ ,..., Rn,
Is determined as the contribution of the I _jth reference data.

Step 4: Using the above-described reference data, feature vectors, and contributions, a stego object is obtained using the method described in Embodiments 4 to 6, and secret data is concealed (embedded).
Step 5: Use the conventional steganography technique to specify index information i ₁ , i at a designated place of a stego object (a place where a pre-designated stego key is embedded and no secret data is embedded). _2, ..., _{i n} a feature vector _{F 1,} F 2, ..., contribution _{a 1, a 2} of _{F n} and _{I j-th} reference data, ..., embedding and _{a n.} That is, the stego key is hidden in the stego object.
Step 6: Send the stego object in which the secret data and the stego key are concealed to the receiving side via the open network.

[Restoring secret data from Stego objects]
On the receiving side, a stego object is acquired via an open network. Then, on the receiving side, the secret data is restored as follows.
Step 1: Index information i ₁ , i ₂ ,..., And _n and feature vectors F ₁ , F _{2 ,.} And contributions a ₁ , a ₂ ,..., _An are extracted.
Step2: index information _{i 1, i 2,} ..., based on _{i n,} select the n pieces of reference data from the reference data set, _{i j-th} feature vector _{F j} of the reference data, the contribution as _{a j,} The secret data is restored using the method described in the fourth to sixth embodiments.

  In the conventional stego object transmission / reception method, the transmission side obtains the stego object from the cover data, then discards the cover data, and the reception side reproduces the secret data based on the stego object and the stego key. In this case, since the security of information completely depends on the stego key, a third party may identify the stego key based on a plurality of stego objects and decrypt the secret information.

On the other hand, in the above-described Stego object transmission / reception method, since the cover data is stored in the local terminal 112 in advance as a reference data set, the cover data need not be sent from the transmission side to the reception side one by one. Therefore, the key information for reproducing the secret data is effectively very large, and even if the existence of the secret information is known, the security of the information becomes very high. Actually, as is clear from [Algorithm for realizing function f for concealing secret data] and [Algorithm for realizing function f ⁻¹ for extracting secret data] described in the sixth embodiment, the secret Data extraction and restoration depend on reference data, transformation of reference data, a combination of reference data, and morphing data. For this reason, it is considered that it is impossible for a third party to specify secret data only by information acquired via an open network by strictly managing reference data.

In the above-described Stego object transmission / reception method, for example, the number n of reference data to be used can be changed every time secret data is different, not a constant. Moreover, the index information i _1, i ₂ of the reference data to be _used, ..., i _n and its contribution a _1, a _2, ..., and a _n, to the receiving side is embedded in stego object with secret data not only send a reception side share the same random number seed as the transmitting side, the index information i ₁ at each location based on the random number seed, i _2, ..., i _n and its contribution a _1, a ₂ , ..., can be generated or calculated and _{a n.} Of course, the index information of the reference data can be obtained not in a random number but in a form depending on the secret data (for example, a numerical value obtained by a hash function or the like based on a sequence of numbers representing the secret data).

  Furthermore, the index information, the feature vector, and the contribution can be sent to the receiving side separately (e-mail) without concealing them in the stego object. Conventionally, the feature vector of the reference data is defined for each transmission. However, the reference data itself may be fixed and stored in the reference data storage device together with the reference data set. If the feature vector is made variable, even if the same reference data is used, if different feature vectors are defined, different morphing data can be generated, so that the safety of information can be further improved. Usually, the amount of feature vectors is much smaller than that of reference data, so that feature vectors can be handled in the same way as index information and contribution.

  In addition, data communication between the transmission side and the reception side can be easily extended to inter-group communication (for example, communication by a social network or the like). When each group member can be a transmission side and a reception side, it is easy to share a common reference data set among group members.

  Furthermore, it can be used not only for communication between the reception side and the transmission side but also for communication between the server and the client as described above. For example, the server generates all reference data and stores it as a reference data set. Then, the server sends a part of reference data (a subset of the reference data set) to a client that is set in advance (for example, a contract for transmission / reception of secret data). When the server wants to send information to a specific client, a stego object may be created using the reference data unique to that client. By transmitting the stego object created by the server, each client can extract information based on the reference data unique to each client.

  Also, instead of normal transmission and reception, the server can embed secret information in a stego object and then post it on a bulletin board (homepage, blog, etc.). A related person (member) can download the stego object from the bulletin board and extract the secret information based on the unique reference data.

  Furthermore, it is possible to send and receive electronic mail using a communication method based on a bulletin board. In this case, since the recipient can receive the mail from the bulletin board as needed instead of passively receiving it, it is possible to avoid spam mail, malicious mail, and the like.

  FIG. 25 and FIG. 26 are diagrams showing an example of sequence processing when mutual authentication and data transmission / reception are performed between the server, the local terminal, and the user's portable device. Specifically, FIG. 25 shows a sequence of preparatory work before performing authentication processing, and FIG. 26 shows a sequence of authentication processing when authentication is actually performed.

  As a system configuration, for example, it is assumed that the user confirms by referring to the secret data whether or not the user is a valid user (holder) authorized by the store side to use. A user carries an IC card or a smart phone or mobile phone having an IC card function (for example, a smart phone or mobile phone having an electronic money function) as a mobile device, and places the mobile device to an operator who operates the local terminal. By presenting it, it is possible to perform authentication work. It is assumed that the local terminal and the server are connected via a network.

Next, preparatory work before performing the authentication process will be described according to the sequence procedure shown in FIG.
Step 1: The server generates a reference data set, encrypts the generated reference data set, and stores it in the reference data storage device of the server. Here, the encryption / decryption key is managed by a server administrator or the like. The encryption / decryption key may be variable.
Step 2: The terminal membership application website is opened by causing the server to function as a web server.
Step 3: Furthermore, the server opens a user application website.
Step 4: On the local terminal, submit the information necessary for membership application (identification code, model, IP address, location status, affiliation, administrator, user, public encryption key, etc.) to the server through the terminal membership application website. To do.

  Step 5: If the server reviews the application information of the local terminal applied through the website for application for terminal registration, the person in charge etc. examines the contents of the information necessary for the application for membership, and the reliability of the terminal for which the application procedure has been performed is low If it can be determined, it is possible to request additional information from the server side to the local terminal, or to reject the application for joining the local terminal. When it is determined that the reliability can be ensured by the information acquired from the local terminal, the local terminal information is additionally registered in the “terminal table” of the server. It is assumed that the contents of the terminal table are stored after being encrypted in the server. The encryption / decryption key for encrypting / decrypting the contents of the terminal table is also managed by the server administrator or the like. The encryption / decryption key may be variable.

Step 6: The server replaces and converts the reference data set, encrypts it with the public key of the local terminal, and transmits it to the local terminal. Further, the server registers the reverse permutation conversion information in the terminal table.
Step 7: The local terminal stores the received reference data set in the reference data storage device on the local terminal side in an encrypted form.
Step 8: The user submits information necessary for member application (ID number, authentication information such as face photo, occupation, work place, contact method, etc.) to the server through the user application website.

  Step9: When the server examines the application information applied through the user application website, and it can be judged that the reliability of the member who has performed the application procedure is low based on the information required for member application Can request additional information from the user or reject the member application. When it is determined that the reliability can be secured by the information acquired from the user, the user information is additionally registered in the “user table” of the server. It is assumed that the contents of the user table are encrypted after being stored in the server. The encryption / decryption key for encrypting / decrypting the contents of the user table is managed by a server administrator or the like. The encryption / decryption key may be variable.

Step 10: The server selects a plurality of reference data from the reference data set, defines the feature vector and contribution of the corresponding reference data, and obtains a stego object that hides the user authentication information.
Step 11: The server divides the stego object and stego key (index, feature vector, contribution, etc.) into two parts (α part and β part) (the way of division is not limited to one method, and various methods are used) Save the divided α part in the user table of the server.
Step 12: Also, the server encrypts the divided β part and transmits it to the user. For this transmission, e-mail can be used as an example. A decryption key for decrypting the encrypted β part is also registered in the user table by the server.
Step 13: The user saves the key information (β portion) sent from the server in the designated area of the portable device while keeping the encryption.

  Authentication preparation work is performed by the processing of Steps 1 to 13 in FIG. 25 described above. By performing the preparatory work in this way, it becomes possible to perform authentication processing of the user's portable terminal and local terminal based on the sequence shown in FIG.

Next, a processing procedure of an authentication process between the local terminal and the user's portable device for the server will be described with reference to the sequence flow shown in FIG.
Step 1: A user presents a mobile device to a local terminal. The local terminal is provided with a read / write device for reading / writing necessary information from a portable device. Data can be transmitted and received via this read / write device.

  Step 2: Mutual authentication is performed between the local terminal and the mobile device. As a mutual authentication method, it is possible to use an authentication technique that has been generally used in credit cards and the like. When it is determined that one of the local terminal and the mobile device is not valid, another method is used for reconfirmation or the service is interrupted. If both are valid, the authentication process after Step 3 is continued.

  Step 3: Local terminal information of the local terminal (identification code, model, IP address, arrangement status, affiliation, administrator, user, public encryption key, etc.) and user information acquired via the read / write device ( The ID number, occupation, work place, contact method, etc.) are encrypted with the public key of the server and transmitted to the server to request key information. It is assumed that the server public key can be easily obtained at the local terminal.

  Step 4: The server decrypts the local terminal and the user information from the received information with a secret key, and confirms the validity of the received information. If either the local terminal or the mobile device is not valid, the server confirms the addition to the local terminal or interrupts the service. If both are valid, the authentication process after Step 5 is continued.

Step 5: Based on the user information, the server acquires from the user table, from the user table, the α part and the decryption key of the β part of the key information for reproducing the user authentication information.
Step 6: The server encrypts the key information acquired from the user table using the public key of the local terminal and sends it to the terminal.
Step 7: The local terminal decrypts the information received from the server with the secret key.
Step 8: The local terminal obtains the β part (encrypted) of the key information (stego key) for reproducing the user authentication information from the portable device, and the obtained part is the decryption key obtained from the server. Decrypt using.
Step 9: The local terminal reproduces the user authentication information by combining the key information (α and β) for reproducing the user authentication information and the data recorded in the reference data storage device.
Step 10: Based on the reproduced user authentication information, the operator authenticates the user.

  By performing authentication processing in this way, it is possible to perform highly secure authentication, and more effectively use the secret data concealment processing and restoration processing shown in the first to sixth embodiments. It becomes possible.

  As described above, the secret information concealment device, the secret information restoration device, the secret information concealment program, and the secret information restoration program according to the present invention have been described in detail by performing case classification in the first to sixth embodiments. The secret information concealment device, secret information restoration device, secret information concealment program, and secret information restoration program according to the present invention are not limited to the examples described in the first to sixth embodiments. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

DESCRIPTION OF SYMBOLS 1 ... Secret information concealment apparatus 2 ... Secret information restoration apparatus 3 ... Network 4 ... Server 6 ... Apparatus body 7 ... Display part 8 ... Data input part 9 ... Command input part 10 ... Algorithm execution part 11 ... Communication part 12 ... Auxiliary storage part 15 ... Calculation part (calculation means)
16 ROM
17 ... RAM
18 ... Storage parts 21, 31, 43, 61, 71, 81, 91 ... Data transformation part (data transformation means)
22, 32, 44, 72, 82, 92... F ₀ generator (F ₀ generator)
23... I ₀ generator (I ₀ generator)
74: Stego object generation unit (stego object generation means)
24, 34, 47, 64, 75, 88, 98, 103 ... Secret data storage units 25, 35, 48, 65, 76, 85, 95 ... Cover data storage units 26, 36, 49, 66, 77, 86, 96 ... Stego key storage unit 27, 37, 50, 67, 78, 87, 97, 101 ... Stego object storage unit 28, 38, 51, 68, 79, 89, 99 ... Working memory 33, 63, 84, 94, 102 ... Secret data reproduction unit (secret data reproduction means)
41 ... Scaling factor calculation unit (scaling factor calculation means)
42, 62 ... Contribution calculation unit (contribution calculation means)
46 ... Morphing data generator (morphing data generator)
73, 83 ... substantial cover data generation unit (substantial cover data generation means)
DESCRIPTION OF SYMBOLS 110 ... Authentication system 111 ... Remote server 112 ... Local terminal 112a ... Card reader 113 ... Card A, B ... Computer

Claims (20)

For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _n, different By applying a morphing technique for generating morphing data based on ν pieces (ν> n) of secret data m _{1 to} m _{ν obtained} by quantifying secret elements of the same type in the same format and a scaling factor A,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key The feature vector {F ₁ , F ₂ ,..., F _n }, the contribution {a ₁ , a ₂ ,..., a _n }, the scaling factor A, and the number of secret data ν are set as k _s. ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
And scale the secret data m _q (q = 1, 2,..., Ν),
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
a stego object generating means for generating a stego object s in which ν secret data m is concealed;
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and scaling factor A secret information hiding apparatus characterized by a positive real number less than 1. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν and the scaling factor A,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, a scaling factor A, and the number of secret data ν, and the extracted secret data m is designated as secret data {m ₁ , m _{2 ,.} As
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
Is used to find the scaled secret data x _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. The process proceeds to the process of determining the above, and the determination process is repeatedly executed until q ≧ ν,
When q <ν is not satisfied,
By extracting ν secret data m _i (i = 1, 2,..., Ν) based on
Secret data reproduction means for extracting ν secret data m concealed in the stego object s,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and scaling factor A secret information restoring apparatus characterized by a positive real number less than 1. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Data {m ₁ , m ₂ ,..., M _n } expressed as numerical values as the secret data m is set, and reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c. The feature vector {F ₁ , F ₂ ,..., F _n } is set as the stego key k _s , the stego object s to be generated is the morphing data I _0, and the feature vector of the morphing data I ₀ is F _0. And the scaling factor is A,
A scaling factor calculation means for obtaining a scaling factor A based on
a _i = m _i × A (values 1 to n are sequentially substituted for i)
, The contribution degree a _i (i = 1, 2,..., N) of each reference data I _i (i = 1, 2,..., N) is converted to the secret data m _i (i = 1, 2, n). .., N) a contribution degree calculation means determined based on
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on
By obtaining the morphing data I ₀ based on
Morphing data generating means for generating morphing data that conceals data expressed numerically with respect to reference data, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , _{respectively,} I _^0, I w 1, · · ^·, showed a p-th element of ^I _{w n,} further scaling factor a secret information hiding apparatus characterized by a positive real number less than 1. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information restoration device for restoring secret data from a stego object s,
Set the morphing data _{I 0} as stego object s, cover reference data as the data _{c {I 1, I 2,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } are set, and the extracted secret data m is set to n pieces of data {m ₁ , m ₂ ,... M _n } represented by numerical values, and the feature vector of the morphing data I ₀ Is F ₀ and the scaling factor is A,
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on
When
Based on these two equations, the solution of the simultaneous equations is obtained, and the contribution a _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N). ) For calculating the degree of contribution
Based on the calculated contribution a _j (j = 1, 2,..., N),
To extract the secret data m _j (j = 1, 2,..., N),
A secret data reproduction means for extracting the secret data concealed in the reference data from the morphing data;
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , _{respectively,} I _^0, I w 1, · · ^·, showed a p-th element of ^I _{w n,} further scaling factor a is secret reconstruction apparatus characterized by a positive real number less than 1. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Set the reference data _{I 1} the secret data m, the cover reference data as the data _{c {I 2, I 3,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the generated stego object s is set as morphing data I ₀ .
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for transforming the secret data and the reference data to obtain I ^w _j (j = 1, 2,..., N) based on
By obtaining the morphing data (stego object) I ₀ based on
I ₀ generating means for concealing secret data from a plurality (three or more) of reference data,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , ^{_{respectively, I 0, I w 1,}} ···, showed a p-th element of ^I _{w n,} Furthermore, the value of the sum of contribution _are those satisfying _{a 1 + a 2 + ··· +} a n = 1 A secret information concealment device characterized by: For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information restoration device for restoring secret data from a stego object s,
Morphing data I ₀ is set as a stego object s, reference data {I ₂ , I ₃ ,..., I _n } is set as cover data c, and feature vectors {F ₁ , F ₂ are set as stego keys k _s. ,..., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the extracted secret data m is set as reference data I ₁ .
F ₀ generating means for obtaining a feature vector F ₀ of the morphing data I ₀ based on
A data transformation means for obtaining transformed data I ^w _j (j = 2, 3,..., N) of each reference data I _j (j = 2, 3,..., N) based on
To obtain the modified data I ^w ₁ of the secret data I ₁ ,
By extracting the secret data I ₁ based on
Secret data reproduction means for extracting secret data concealed with respect to a plurality of reference data,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , ^{_{respectively, I 0, I w 1,}} ···, showed a p-th element of ^I _{w n,} Furthermore, the value of the sum of contribution _are those satisfying _{a 1 + a 2 + ··· +} a n = 1 A secret information restoration apparatus characterized by: For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
s _p = MSB (I ⁰ _0p ) + m _q
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
a stego object generating means for generating a stego object s in which ν secret data are concealed;
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and MSB (I ⁰ _0p ) is the upper B− of the I ⁰ _0p. A secret information concealment device, which means a function for extracting b-bit data. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
_{Δ p (j 1, j 2} ) = (I w j1, p -I w j2, p) if and ≠ 0 become _{j 1} and _{j 2} are present, m based on the corresponding _{j 1} and _{j 2 q} = s _p -MSB (s _p )
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
Secret data reproduction means for extracting ν secret data concealed in the stego object,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1,2, ···, ν ) is the data of b ^bits, _{I 0 0p} is a B data bits, MSB _{(s p),} the upper B-b bits of _{s p} This means a function for retrieving the data of the secret information. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment device for generating a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
Based on the above, the substantial cover data generating means for obtaining the morphing data I ⁰ ₀ before concealing the secret data,
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Seeking stego object s _p was used to increase the value of p by 1, further increasing the value of q by 1,
After that, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the determination process and repeatedly executing the determination process until q ≧ ν,
a stego object generating means for generating a stego object s in which ν secret data are concealed;
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and MSB (I ⁰ _0p ) is the upper B− of the I ⁰ _0p. means a function that retrieves b-bit data;
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits A secret information concealment device characterized in that numerical data is set. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
F ₀ generating means for obtaining a feature vector F ₀ of morphing data using
A data transformation means for obtaining transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using
After p is set to 1 and q is set to 1 and variables p and q are initialized,
It is determined whether j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) a ≠ 0 _{j 1} and _{j 2} and is not present, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
After that, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
Secret data reproduction means for extracting ν secret data concealed in the stego object,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 to satisfy the, also, secret data _{m q} q = 1,2, ···, ν) is the data of b ^bits, a data _{I 0 0p} is B bits, MSB _{(s p),} the upper B-b bits of _{s p} Means a function that retrieves data, and
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits A secret information restoration device characterized in that numerical data is set. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _n, different By applying a morphing technique for generating morphing data based on ν pieces (ν> n) of secret data m _{1 to} m _{ν obtained} by quantifying secret elements of the same type in the same format and a scaling factor A,
A secret information concealment program for a secret information concealment device that generates a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key The feature vector {F ₁ , F ₂ ,..., F _n }, the contribution {a ₁ , a ₂ ,..., a _n }, the scaling factor A, and the number of secret data ν are set as k _s. ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
And scale the secret data m _q (q = 1, 2,..., Ν),
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
With seeking stego object s _p, the value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
a stego object generation function for generating a stego object s in which ν secret data m is concealed;
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and scaling factor Secret information hiding program secret information hiding apparatus A which is characterized in that less than 1 positive real number. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν and the scaling factor A,
A secret information restoration program of a secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, a scaling factor A, and the number of secret data ν, and the extracted secret data m is designated as secret data {m ₁ , m _{2 ,.} As
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To obtain the scaled secret data x _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. The process is shifted to the process of determining whether the determination process is repeated until q ≧ ν,
When q <ν is not satisfied,
By extracting ν secret data m _i (i = 1, 2,..., Ν) based on
A secret data reproduction function that extracts ν secret data m concealed in the stego object s, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and scaling factor Secret reconstruction program secret reconstruction device A which is characterized in that less than 1 positive real number. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information concealment program for a secret information concealment device that generates a stego object s in which secret data is concealed,
Data {m ₁ , m ₂ ,..., M _n } expressed as numerical values as the secret data m is set, and reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c. The feature vector {F ₁ , F ₂ ,..., F _n } is set as the stego key k _s , the stego object s to be generated is the morphing data I _0, and the feature vector of the morphing data I ₀ is F _0. And the scaling factor is A,
In the calculation means of the secret information concealment device,
A scaling factor calculation function for obtaining a scaling factor A based on
a _i = m _i × A (values 1 to n are sequentially substituted for i)
, The contribution degree a _i (i = 1, 2,..., N) of each reference data I _i (i = 1, 2,..., N) is converted to the secret data m _i (i = 1, 2, n). .., N) a contribution calculation function to be determined based on
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on ,
By obtaining the morphing data I ₀ based on
A morphing data generation function that generates morphing data that conceals data expressed numerically with respect to reference data, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , _{respectively,} I _^0, I w 1, · · ^·, showed a p-th element of ^I _{w n,} further secret scaling factor a secret information hiding apparatus characterized by a positive real number less than 1 Information hiding program. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, on the basis of the n data _m 1 ~m _n being represented by a numerical value, morphing data I _0, and the feature vector F ₀ morphing data I _0, by applying a morphing technique for generating a scaling factor a,
A secret information restoration program of a secret information restoration device for restoring secret data from a stego object s,
Set the morphing data _{I 0} as stego object s, cover reference data as the data _{c {I 1, I 2,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } are set, and the extracted secret data m is set to n pieces of data {m ₁ , m ₂ ,... M _n } represented by numerical values, and the feature vector of the morphing data I ₀ Is F ₀ and the scaling factor is A,
In the calculation means of the secret information restoration device
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) based on ,
When
Based on these two equations, the solution of the simultaneous equations is obtained, and the contribution a _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N). )
Based on the calculated contribution a _j (j = 1, 2,..., N),
To extract the secret data m _j (j = 1, 2,..., N),
Realize the secret data reproduction function to extract the secret data hidden in the reference data from the morphing data,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , _{respectively,} I _^0, I w 1, · · ^·, showed a p-th element of ^I _{w n,} further secret scaling factor a is secret reconstruction apparatus characterized by a positive real number less than 1 Information restoration program. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information concealment program for a secret information concealment device that generates a stego object s in which secret data is concealed,
Set the reference data _{I 1} the secret data m, the cover reference data as the data _{c {I 2, I 3,} ···, I n} is set and the feature vector _{F 1 as stego key _{k s, F} 2, .., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the generated stego object s is set as morphing data I ₀ .
In the calculation means of the secret information concealment device
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
And a data transformation function for transforming the secret data and the reference data to obtain I ^w _j (j = 1, 2,..., N),
By determining the morphing data (stego object) I ₀ based on
An I ₀ generation function for concealing secret data from multiple (three or more) reference data, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , ^{_{respectively, I 0, I w 1,}} ···, showed a p-th element of ^I _{w n,} Furthermore, the value of the sum of contribution _are those satisfying _{a 1 + a 2 + ··· +} a n = 1 A secret information concealment program for a secret information concealment device. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and reference data _I 1 ~I _n of n (n ≧ 3), feature vector _F 1 to F _n of the reference data _I 1 ~I _n, contribution _a 1 ~ in each of the reference data _I 1 ~I _n based on the a _n, by applying a morphing technique for generating morphing data I _0,
A secret information restoration program of a secret information restoration device for restoring secret data from a stego object s,
Morphing data I ₀ is set as a stego object s, reference data {I ₂ , I ₃ ,..., I _n } is set as cover data c, and feature vectors {F ₁ , F ₂ are set as stego keys k _s. ,..., F _n } and contributions {a ₁ , a ₂ ,..., A _n } are set, and the extracted secret data m is set as reference data I ₁ .
In the calculation means of the secret information restoration device,
F ₀ generation function for determining the feature vector F ₀ of the morphing data I ₀ based on
A data transformation function for obtaining the transformed data I ^w _j (j = 2, 3,..., N) of each reference data I _j (j = 2, 3,..., N) based on ,
To obtain the modified data I ^w ₁ of the secret data I ₁ ,
By extracting the secret data I ₁ based on
A secret data reproduction function that extracts secret data hidden from multiple reference data,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, I 0p, I w 1p,}} ···, I w np is , ^{_{respectively, I 0, I w 1,}} ···, showed a p-th element of ^I _{w n,} Furthermore, the value of the sum of contribution _are those satisfying _{a 1 + a 2 + ··· +} a n = 1 A secret information restoration program for a secret information restoration device. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment program for a secret information concealment device that generates a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
s _p = MSB (I ⁰ _0p ) + m _q
With seeking stego object s _p, the value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
a stego object generation function for generating a stego object s in which ν secret data are concealed, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and MSB (I ⁰ _0p ) is the upper B− of the I ⁰ _0p. A secret information concealment program for a secret information concealment apparatus, which means a function for extracting b-bit data. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration program of a secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
_{Δ p (j 1, j 2} ) = (I w j1, p -I w j2, p) if and ≠ 0 become _{j 1} and _{j 2} are present, m based on the corresponding _{j 1} and _{j 2 q} = s _p -MSB (s _p )
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining the above and repeatedly executing the determination process until q ≧ ν,
Realize a secret data reproduction function that extracts ν secret data concealed in a stego object,
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1,2, ···, ν ) is the data of b ^bits, _{I 0 0p} is a B data bits, MSB _{(s p),} the upper B-b bits of _{s p} A secret information restoration program for a secret information restoration device, characterized in that it means a function for retrieving the data. For the steganography technique for generating the stego object s by concealing the secret data m in the cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces of secret data m _{1 to} m _v ,
A secret information concealment program for a secret information concealment device that generates a stego object s in which secret data is concealed,
Secret data {m ₁ , m ₂ ,..., M _v } is set as the secret data m, reference data {I ₁ , I ₂ ,..., I _n } is set as the cover data c, and the stego key Set feature vectors {F ₁ , F ₂ ,..., F _n } and contributions {a ₁ , a ₂ ,..., a _n } as k _s ,
In the calculation means of the secret information concealment device,
An F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
A substantial cover data generation function for obtaining morphing data I ⁰ ₀ before concealing secret data based on
After setting the value of 1 to p, the value of 1 to q, the value of I ⁰ ₀ to s, and initializing the variables p, q and s,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p of the _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} are present, based on the relevant _{j 1} and _{j 2} ,
Let me seek stego object s _p with a value of p is increased by 1, and further increases the value of q by 1,
Then, determine whether q <ν is satisfied,
Whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 when q <ν is satisfied is determined. By shifting to the process of determining and repeatedly executing the determination process until q ≧ ν,
a stego object generation function for generating a stego object s in which ν secret data are concealed, and
Where F _jp (j = 0, 1,..., N) indicates the p-th element of F _j , and I ^w _j (j = 1, 2,..., N) is a modification of I _j W denotes a deformation (warping) function, and I _j (j = 1, 2,..., N) is deformed by the deformation function W, so that I ^w _j (j = 1, 2,... , n) is _{^{generated, I w j (j = 1,2}} , ···, n) has a feature vector _{F 0, ^{also, s p, I 0 0p,}} I w 1p, ···, I ^w _np represents the p-th element of s, I ⁰ ₀ , I ^w ₁ ,..., I ^w _n , respectively, and j ₁ and j ₂ are 1 ≦ j ₁ <j ₂ ≦ is a natural number satisfying n, contribution _{a j (j = 1,2, ···} , n) value of the previously normalized value of the sum of the contributions _{is, a 1 + a 2 + ···} + a n = 1 and the secret data m _q (Q = 1, 2,..., Ν) is b-bit data, I ⁰ _0p is B-bit data, and MSB (I ⁰ _0p ) is the upper B− of the I ⁰ _0p. means a function that retrieves b-bit data;
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits A secret information concealment program for a secret information concealment device, characterized in that numerical data is set. For steganography technology that extracts secret data m from stego object s in which secret data m is hidden in cover data c,
and n pieces of reference data _I 1 ~I _n, and the feature vector _F 1 to F _n of the reference data _I 1 ~I _n, and contribution _a 1 ~a _n in each of the reference data _I 1 ~I _{n, ν} By applying a morphing technique for generating morphing data based on the pieces (ν> n) of secret data m _{1 to} m _ν ,
A secret information restoration program of a secret information restoration device for restoring secret data from a stego object s,
Reference data as a cover data _{c {I 1, I 2,} ···, I n} is set and the feature vector as the stego key _{k s {F 1, F 2} , ···, F n} and contribution {a ₁ , a ₂ ,..., A _n }, and the extracted secret data m is set as secret data {m ₁ , m _{2 ,.}
In the calculation means of the secret information restoration device,
F ₀ generation function for obtaining a feature vector F ₀ of morphing data using
And a data transformation function for obtaining the transformed data I ^w _j (j = 1, 2,..., N) of each reference data I _j (j = 1, 2,..., N) using ,
After p is set to 1 and q is set to 1 and variables p and q are initialized,
Whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
If delta _p where _{(j 1, j 2) =} (I w j1, p -I w j2, p) ≠ 0 and becomes _{j 1} and _{j 2} and does not exist, increasing the value of p by 1, It is repeatedly executed whether or not j ₁ and j _{2 satisfying} Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist,
Delta _{p (j 1,} j 2) = ^{when _{(I w j1, p -I w}} j2, p) and _{j 1} and _{j 2} as a ≠ 0 is present, based on the relevant _{j 1} and _{j 2}
To determine the secret data m _q , increase the value of p by 1, and further increase the value of q by 1.
Then, determine whether q <ν is satisfied,
If q <ν is satisfied, whether or not j ₁ and j ₂ satisfying Δ _p (j ₁ , j ₂ ) = (I ^w _{j1, p} −I ^w _{j2, p} ) ≠ 0 exist. By shifting to the process of determining the above and repeatedly executing the determination process until q ≧ ν,
A secret data reproduction function for extracting ν secret data concealed in a stego object is realized, where F _jp (j = 0, 1,..., N) indicates the p th element of F _j , I ^w _j (j = 1, 2,..., N) indicates a deformation of I _j , W indicates a deformation (warping) function, and I _j (j = 1, 2,... ·, n) is the _{^{deformation, I w j (j = 1,2}} , ···, n) is _{^{generated, I w j (j = 1,2}} , ···, n) is the feature vector F has _{0, ^{also, s p, I 0 0p,}} I w 1p, ···, I w np , ^{_{respectively, s, I 0 0, I}} w 1, ···, p th of ^I _{w n} In addition, j ₁ and j ₂ are natural numbers that satisfy 1 ≦ j ₁ <j ₂ ≦ n, and the value of the contribution a _j (j = 1, 2,..., N) is determined in advance. regular The sum of the contributions satisfies a ₁ + a ₂ +... + A _n = 1, and the secret data m _q (q = 1, 2,..., Ν) a data ^to a data _{I 0 0p} is B bits, MSB _{(s p)} denotes a function to extract the upper B-b-bit data of the _{s p,} furthermore,
Indicates an exclusive OR operator in units of bits, and y is either I ⁰ ₀ , I ^w _np or (I ^w _{j1, p} −I ^w _{j2, p} ), and is b bits A secret information restoration program for a secret information restoration device, characterized in that numerical data is set.