CN110430335A - A method, device and storage medium for information camouflage and restoration - Google Patents

Abstract

The invention belongs to the intersection field of image information security and digital image signal processing, and discloses an information camouflage and recovery method, device and storage medium. In camouflage, first encode the secret information into the coordinates on the key image to avoid the direct channel transmission of the secret information, then introduce multiple backups, further encode the coordinate points into the high-frequency color sequence of the given sample image and use the key Pseudo-randomly placed on the blank texture image to be synthesized, and finally a pixel-by-pixel texture generation strategy is used to generate a dense texture image. When recovering, the color sequence is extracted according to the key, and then the coordinates are recovered from the sample image combined with the interval expansion strategy of multiple backups, and the secret information is extracted according to the key image. Compared with existing methods, this method synthesizes texture pixel by pixel without leaving obvious splicing traces and repeated patterns, and the secret information is completely dependent on the key, which has certain anti-attack ability and high security .

Description

A method, device and storage medium for information camouflage and recovery

technical field

The invention belongs to the intersection field of image information security and digital image signal processing, and relates to a method, device and storage medium for information camouflage and restoration.

Background technique

With the development of the Internet, more and more information is transmitted through the network. While bringing great convenience to people, there are also certain security risks. In order to protect the security of secret information in transmission, people propose to use images to hide secret information, for example, for steganographic literature: Mao Binghua, 2019 (Mao Binghua, Wang Zichi, Zhang Xinpeng. Asymmetric JPEG steganography based on DCT domain correlation[J ]. Computer Science, 2019, 46(01): 203-207.); GANDHARBA S, 2016 (GANDHARBAS. Adaptive pixel value differencing steganography using both vertical and horizontal[J]. Multimedia Tools and Applications, 2016, 75(21): 13541-13556.); YANG TY, 2017 (YANG TY, CHEN H S. Matrix embedding in steganography with binary Reed-Muller codes [J]. IET Image Processing, 2017, 11(7): 522-529.) and Zhang Yang , 2018 (Zhang Yang, Shao Liping, Ren Pingan. Basis-free EMD(n, m) model and its application in image steganography [J]. Journal of Computer-Aided Design and Graphics, 2018, 30(8): 1490-1504.); Literature for shared storage: Ouyang Xianbin, 2017 (Ouyang Xianbin, Shao Liping, Le Zhifang. Non-equivalent backup and dual-authentication self-repair finite field image sharing[J]. Software Journal, 2017, 12(28): 3306-3346.) and Huang Yanyan, 2018 (Huang Yanyan, Shao Liping. Multi-carrier dense image sharing method combined with EMD- ^cl embedding [J]. Chinese Journal of Image and Graphics, 2018, 23( 8): 1108-1130.) on. However, the essence of these methods is the traditional modified embedding information hiding method, which needs to modify the carrier image, which will have a great impact on the visual quality of the embedded carrier and easily arouse the suspicion of the attacker.

In order to avoid such problems, texture generative no-carrier information hiding is proposed, which mainly includes texture construction information hiding and texture splicing information hiding. Among them, texture construction information hiding is to generate some natural textures by simulating texture generation, for example: QIAN Z X, 2018 (QIAN Z X, PAN L, LI S, ZHANG XP. Steganography by constructing marbling texture[C]//section //2018ChinaInformation Hiding Workshop (CIHW2018), Guang Zhou, 2018: 17-35.); Pan Lin, 2016 (Pan Lin, Qian Zhenxing, Zhang Xinpeng. Digital steganography based on structured texture images [J]. Chinese Journal of Applied Science, 2016 , 34(5): 625-632.) and XU J, 2015 (XU J, MAO X, JIN X, et al. Hidden message in a deformation-based texture[J]. Visual Computer International Journal of Computer Graphics, 2015, 31(12): 1653-1669.), etc. These methods hide the secret information by generating water-like shadow painting textures, but the generated textures are all unnatural textures, so the secret information cannot be effectively covered.

Texture splicing hiding originated from OTORI H, 2007 (OTORI H, KURIYAMA S. Data-embeddable texture synthesis[C]//International Symposium on Smart Graphics, Kyoto, Japan, 2007: 146-157.) and OTORI H, 2009 (OTORI H, KURIYAMA S. Texture synthesis for mobile data communications[J]. IEEE Computer Graphics and Applications, 2009, 29(6): 74-81.), this type of method expresses secret information by filling sample textures However, since the local binary pattern cannot be well covered, it is easy to cause the suspicion of the steganalyzer and lead to the leakage of secret information. To avoid this problem, WU C, 2015 (WU C, WANG C M. Steganography using reversible texture synthesis[J]. Transactions on Image Processing, 2015, 24(1): 130-139.); QIN Z C, 2017 (QIN Z C , LI M, WU B. Robust steganography viapatch-based texture synthesis[C]//International Conference on InternetMultimedia Computing and Service. Springer, Singapore, 2017: 429-439.) and QIAN Z X, 2016 (QIAN Z X, ZHOU H, ZHANG W M, et al. Robust steganography using texture synthesis[C]//Advances in Intelligent Information Hiding and Multimedia Signal Processing. Proceeding of the Twelfth International Conference on Intelligent Information Hiding and Multimedia Signal Processing, 2016: 21-23. Block concatenation is used to generate dense textures similar to a given sample. However, WU C, 2015 introduced the mirroring operation to make the coded and non-coded sample blocks have obvious distinguishing features, resulting in the direct leakage of coded small blocks. QIN Z C, 2017 and QIAN Z X, 2016 There is a fixed one-to-one correspondence between secret information segments and sample small block categories, which can easily arouse the suspicion of attackers and lead to low security. In addition, WU C, 2015, QIN Z C, 2017 and QIAN Z X, 2016 contain dense texture images using EFROS A A, 2001 (EFROS A A, FREEMAN W T.Image quilting fortexture synthesis and transfer[C]//Proc.of the 28th Annual Conference on Computer Graphics and Interactive Techniques, 2001: 341-346.) stitches textures in overlapping regions, which will inevitably produce stitching traces and repeated texture patterns, so that secret information cannot be concealed.

Contents of the invention

The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art, and provide a method, device and storage medium for disguising and recovering information.

In order to achieve the above object, the present invention adopts the following technical solutions to achieve:

A method for disguising information comprises the following steps:

S1: Generate an R-order key map S through the preset key k ₁ , input a binary secret information bit sequence B, and encode the binary secret information bit sequence B into a 2-ary ^R number sequence R;

S2: Encode the 2-ary number sequence ^R into a binary bit string b through the pixel coordinates of the same pixel value in the R-order key map S as the element in the 2-ary number sequence ^R ;

S3: Select m+t colors from the preset texture sample map T to form a palette sequence V, initialize a blank image K, and generate a random coordinate sequence C through the preset key k ₂ ;

S4: Through the mapping relationship between elements and element indexes in the palette sequence V, encode the 3-fold backup of the binary bit string b into a color sequence V', combine the random coordinate sequence C to place the color sequence V' on the blank image K all elements in

S5: Fill the remaining positions in the blank image K with texture through a pixel-by-pixel texture generation strategy to obtain a dense texture image K.

The further improvement of the information camouflaging method of the present invention lies in:

The specific method of said S1 is:

S1-1: Generate an R-order key map with a resolution of h ₁ ×w ₁ through the preset key k ₁

S1-2: Input the 2-value secret information bit sequence B, convert each R bit in B into a group of 2 ^R -ary number sequences with length l ₁ Wherein l ₁ is determined by formula (1);

Among them, the symbol Indicates rounding up;

The specific method of said S2 is:

Encode the 2-ary number sequence ^R into a 2-valued bit string through the pixel coordinates of the same pixel value in the R-order key map S as the element in the 2-ary number sequence ^R Wherein l ₂ is determined by formula (2);

The specific method of said S2 is:

S2-1: Pair of 2 ^R -ary number sequences For each element r _i in the R-order key map S, randomly select an element equal to the value of r _i and record its corresponding coordinate position (u _i , v _i ), and convert u _i and v _i into lengths of The binary bit string of

Among them, bin(u _i ) is a binary conversion function, which is used to convert u _i into a binary bit string. For u _i ∈ [0, h ₁ -1], bin(u _i ) converts u _i into length for 2-value bit string of , for v _i ∈ [0, w ₁ -1], bin(v _i ) converts v _i to length The 2-value bit string of , the symbol "||" is a bit string concatenation function;

S2-2: Repeat S2-1 to Convert to l ₁ binary bit string Through formula (4) will l ₁ binary bit string Concatenate to get 2-valued bit string

The specific method of said S3 is:

S3-1: Sample images of preset textures at resolution h ₂ ×w ₂ Select the m+t colors with the highest frequency of occurrence to form a palette sequence V=(v _i ) _m+t ;

S3-2: Initialize an R-order blank image with a resolution of h ₃ ×w ₃

S3-3: Generate a random coordinate sequence of length 3 l ₃ through the preset key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in the random coordinate sequence C are not equal, l ₃ is determined by formula (5):

Among them, the symbol Indicates rounding down;

The specific method of said S4 is:

S4-1: Bind the 2-valued bits in the 2-valued bit string b to A group of bits is converted into a sequence of length l ₃

S4-2: Convert the sequence Each element d _i in is encoded as a backup of d _i by formula (6)

S4-3: Repeat S4-2 to encode D as 3 l ₃ elements Obtain the color sequence V' by formula (7);

S4-4: will As the blank image K (x _3·i+k , y _3·i+k ), i=0, 1,..., l ₃ -1, k=0, 1, 2 elements;

The concrete method of described S5 is:

S5-1: In the blank image K, for any k _{i, j} and Denote the γ×γ neighborhood centered on _{ki, j} as N(ki _{, j} ), traverse all the γ×γ neighborhoods N′(t _x y) centered on all elements t _{x, y} in the texture sample map _{, y} ), get the similarity between N(ki _{, j} ) and N′(t _{x, y} ) through formula (8), and select the neighborhood most similar to N(ki _{, j} ) according to Assign values to k _{i, j} ;

Among them, p, q are the corresponding pixel points in the 8 neighborhoods N(k _{i, j} ), N′(t _{x, y} ), and R(x), G(x), B(x) respectively represent the pixels For the R, G, and B component values of point x, the smaller d(N(k _{i, j} ), N′(t _{x, y} )), the smaller N(k _{i, j} ) and N′(t _{x, y} ) are. resemblance;

S5-2: Repeat S5-1 to all elements in the blank image K After completing the assignment, a dense texture image K is obtained.

In yet another aspect of the present invention, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the information masquerading method are realized.

In yet another aspect of the present invention, a computer device includes a memory, a processor, and a computer program stored in the memory and operable on the processor, and the information is realized when the processor executes the computer program. The steps of the masquerading method.

In yet another aspect of the present invention, an information recovery method includes the following steps:

R1: Generate a random coordinate sequence C through the preset key k ₂ , and extract the color sequence V' from the input dense texture map K' through the random coordinate sequence C;

R2: Select m+t colors from the preset texture sample map T to form a palette sequence V, and decode the color sequence V' into 2 values through the mapping relationship between the elements in the palette sequence V and the element indexes bit string b;

R3: Generate the R-order key map S through the preset key k ₁ , and extract the secret information B from the key map S by combining the binary bit string b.

The further improvement of the information recovery method of the present invention lies in:

The concrete method of described R1 is:

R1-1: Generate a random coordinate sequence with a length of 3 l ₃ through the preset key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in the random coordinate sequence C are not equal to each other;

R1-2: From the input dense texture map through the random coordinate sequence C , for each (x _i , y _i )∈C, find (x _3·i+k , y _3·i+k ), i∈[0, l ₃ -1], k=0, 1, 2 all elements of position and combined to obtain the color sequence V';

The concrete method of described R2 is:

R2-1: Texture sample image at resolution h ₂ ×w ₂ Select the m+t colors with the highest frequency of occurrence to form a palette sequence V=(v _i ) _m+t ;

R2-2: Find out according to the palette sequence V index position of Calculated by formula (9)

remember The number of elements in the interval [0, m-1] is recorded as E _c , and d _i is calculated through the following five situations:

①If E _c =0, set

②If E _c =1 and The element value in the interval [0, m-1] is v, set d _i =v;

③If E _c =2 and The element values in the interval [0, m-1] are v ₁ , v ₂ , set d _i =(v ₁ +v ₂ )/2;

④ If E _c =3 and There are two or more values that are equal, record the equal value as v, and set d _i =v;

⑤If E _c =3 and Not equal to each other, set

R2-3: Repeat R2-2 until all the coordinates in the random coordinate sequence C have been processed, and the ₂ -value bit string b of length l2 is obtained by formula (10):

Wherein, Left () is a binary bit string interception function, the first parameter is the binary bit string to be intercepted, and the second parameter is the length intercepted from the left; ₁₂ is determined by formula (11);

The concrete method of described R3 is:

R3-1: Generate an R-order key map with a resolution of h ₁ ×w ₁ through the preset key k ₁

R3-2: Bind the 2-valued bits of the 2-valued bit string b to is 1 group, divided into l ₁ groups, recorded as From each group through formula (12) and formula (13) respectively before interception 2-valued bits and after two 2-value bits as the coordinates (u _i , v _i ), and then assign the element corresponding to the position (u _i , v _i ) in the R-order key map S to r _i through formula (14), where i=0, 1 ,...,l ₁ -1;

Wherein, Right () is a binary bit string interception function, the first parameter is the binary bit string to be intercepted, and the length of the second parameter intercepted from the right;

R3-3: convert r _i , i=0, 1, ..., l ₁ -1 into binary bit strings of length R respectively, and concatenate all binary bit strings of length R into a length of The binary bit string of N, the secret information B=(B _i ) _N is obtained.

In yet another aspect of the present invention, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the information restoration method are implemented.

In yet another aspect of the present invention, a computer device includes a memory, a processor, and a computer program stored in the memory and operable on the processor, and the information is realized when the processor executes the computer program. The steps of the recovery method.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention adopts the pixel-by-pixel texture synthesis method to generate dense texture, and the dense texture image is generated point by point instead of splicing small texture blocks, so it avoids the inherent splicing existing in the block splicing generative information hiding based on the suture line algorithm Traces and reduce the generation of repeated patterns, higher security, and less likely to arouse the suspicion of attackers. It effectively solves the problem that in the traditional information hiding method based on block splicing texture synthesis, because the dense texture image uses the seam algorithm to stitch the texture of the overlapping area, it will inevitably produce stitching traces and repeated texture patterns, so that the secret information cannot be hidden. cover up the problem. The secret information is expressed through the coordinates of the key map generated by the key, and the coordinate information of the secret information in the key map is embedded in the secret bunker map instead of the secret information, which avoids the direct channel transmission of the secret information, and the secret information The process of embedding and extracting completely depends on the key, which has high security. It effectively solves the problem that the existing information hiding method based on block splicing texture synthesis is not high in security due to the use of small sample blocks to encode secret information, and the secret information is transmitted in the channel. The present invention introduces an authentication strategy based on multi-backup and interval expansion. When embedding multiple secret information, when extracting, all multiple secret information are extracted, and then compared with multiple secret information, the secret information is the same at most The secret information is taken as the correctly extracted secret information, and if they are not the same, the average value is taken as the secret information. This strategy can effectively improve the anti-attack ability of extracting secret information. It effectively solves the problem that the existing information hiding method based on block splicing texture synthesis lacks authentication information, and it is impossible to determine whether the extracted secret information is correct secret information if the secret bunker map is attacked.

Description of drawings

Fig. 1 is the flow chart of information camouflage method of the present invention;

Fig. 2 is a flow chart of the information recovery method of the present invention;

Fig. 3 is a texture sample diagram of the present invention, which is an 8-bit color image texture 1 with a resolution of 64×64;

Fig. 4 is a texture sample diagram of the present invention, which is an 8-bit color image texture 2 with a resolution of 64 × 64;

Fig. 5 is a texture sample diagram of the present invention, which is an 8-bit color image texture 3 with a resolution of 64×64;

Fig. 6 is a texture sample diagram of the present invention, which is an 8-bit color image texture 4 with a resolution of 64×64;

Fig. 7 is a dense image of the present invention, which is an 8-bit grayscale image lena of 64 × 64 resolution;

Fig. 8 is a sample map of the texture in Fig. 3 of the present invention, and Fig. 7 is a dense texture 1 with a resolution of 512 × 512 generated by the dense map, and the user key is key ₁ = 6546, key ₂ = 7653;

Fig. 9 is a sample map of the texture in Fig. 4 of the present invention, Fig. 7 is a dense texture 2 with a resolution of 512 × 512 generated by the dense map, and the user key is key ₁ = 6548, key ₂ = 7659;

Fig. 10 is an example map of the texture in Fig. 5 of the present invention, and Fig. 7 is a dense texture 3 with a resolution of 512 × 512 generated by the dense map, and the user key is key ₁ = 6546, key ₂ = 7653;

Fig. 11 is an example map of the texture in Fig. 6 of the present invention, and Fig. 7 is a dense texture 4 with a resolution of 512 × 512 generated by the dense map, and the user key is key ₁ = 6548, key ₂ = 7659;

Fig. 12 is the secret map recovered by Fig. 8 of the present invention, and the bit error rate BR is 0% with respect to Fig. 7;

Fig. 13 is the secret map recovered by Fig. 9 of the present invention, and the bit error rate BR relative to Fig. 7 is 0%;

Fig. 14 is the cryptography restored by Fig. 10 of the present invention, and is 0% with respect to the bit error rate BR of Fig. 7;

Fig. 15 is the secret map restored by Fig. 11 of the present invention, and is 0% with respect to the bit error rate BR of Fig. 7;

Fig. 16 is a 1.9% clipping attack diagram for Fig. 8 of the present invention;

Fig. 17 is a 3.4% clipping attack diagram of Fig. 9 in the present invention;

Figure 18 is a 4.1% clipping attack diagram of Figure 10 of the present invention;

Fig. 19 is a 2.6% clipping attack diagram of Fig. 11 according to the present invention;

Fig. 20 is the 8% random noise attack diagram of Fig. 8 of the present invention;

Figure 21 is a 20% random noise attack diagram of Figure 9 of the present invention;

Figure 22 is a 15% random noise attack figure for Figure 10 of the present invention;

Figure 23 is a 10% random noise attack diagram of Figure 11 of the present invention;

Fig. 24 is the encrypted image recovered from Fig. 16 according to the present invention. Compared with Fig. 7, the bit error rate BR is 0.63%, and the peak signal-to-noise ratio PSNR=is 30.7281dB;

Fig. 25 is the encrypted image recovered from Fig. 17 according to the present invention. Compared with Fig. 7, the bit error rate BR is 1.12%, and the peak signal-to-noise ratio PSNR=is 28.1291dB;

Fig. 26 is the encrypted image recovered from Fig. 18 according to the present invention. Compared with Fig. 7, the bit error rate BR is 1.37%, and the peak signal-to-noise ratio PSNR=26.7750dB;

Fig. 27 is the encrypted image recovered from Fig. 19 according to the present invention. Compared with Fig. 7, the bit error rate BR is 0.85%, and the peak signal-to-noise ratio PSNR=29.8146dB;

Fig. 28 is the encrypted image recovered from Fig. 20 according to the present invention. Compared with Fig. 7, the bit error rate BR is 2.66%, and the peak signal-to-noise ratio PSNR=24.3136dB;

Fig. 29 is the encrypted image recovered from Fig. 21 according to the present invention. Compared with Fig. 7, the bit error rate BR is 6.35%, and the peak signal-to-noise ratio PSNR=20.1874dB;

Fig. 30 is the encrypted image recovered from Fig. 22 according to the present invention. Compared with Fig. 7, the bit error rate BR is 4.79%, and the peak signal-to-noise ratio PSNR=21.7039dB;

Fig. 31 is the encrypted image recovered from Fig. 23 according to the present invention. Compared with Fig. 7, the bit error rate BR is 3.13%, and the peak signal-to-noise ratio PSNR=23.7018dB.

Detailed ways

In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only It is an embodiment of a part of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

It should be noted that the terms "first" and "second" in the description and claims of the present invention and the above drawings are used to distinguish similar objects, but not necessarily used to describe a specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the invention described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to the expressly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

The present invention will be further described in detail below with JAVA jdk1.8.0_65 as the implementation environment in conjunction with the accompanying drawings and embodiments.

The present invention is described in further detail below in conjunction with accompanying drawing:

Referring to Fig. 1, the information camouflage method of the present invention comprises the following steps:

Step 1: First generate an R-order key map with a resolution of h ₁ ×w ₁ through the preset key k ₁ Then input a binary secret information bit sequence B=(B _i ) _N with a length of N, and finally convert each R bit in the binary secret information bit sequence B into a group of 2 ^R -ary numbers with a length of l ₁ sequence Wherein l ₁ is determined by formula (1);

Among them, the symbol Indicates rounding up.

For example: if the random seed of the linear congruent random generator is used as the key k ₁ , then S=(s _{i, j} ) _16×16 can be generated by k ₁ =6546, R=8, h ₁ =w ₁ =16 , s _{i, j} ∈ [0, 2 ⁸ -1], assuming the input binary secret information bit sequence B=(1101011010110010) ₂ , then N=16, then the length can be obtained according to formula (1): 2 ^R =2 ⁸ =256 hexadecimal number sequence R, R contains 2 elements r ₀ =(11010110) ₂ =214, r ₁ =(10110010) ₂ =178, namely R=(r ₀ =214, r ₁ = 178).

Step 2: For 2 ^R -ary number sequences For each element r _i in the R-order key map S, randomly select an element equal to the value of r _i And record its corresponding coordinate position (u _i , v _i ), convert u _i and v _i into lengths according to formula (3) The binary bit string of From this you can Convert to l ₁ binary bit string Concatenate the bits according to formula (4) to get the 2-value bit string Wherein l ₂ is determined by formula (2);

In formula (3), bin(u _i ) is a binary conversion function, which is used to convert u _i into a binary bit string. For u _i ∈ [0, h ₁ -1], bin(u _i ) can be U _i is converted to a length of 2-value bit string of , for v _i ∈ [0, w ₁ -1], bin(v _i ) can convert v _i into a length of The 2-value bit string of , the symbol "||" is a bit string concatenation function;

For example: for 2 elements r ₀ =214, r ₁ =178 in R=(r ₀ =214, r ₁ =178), if the elements in the R-order key map S equal to r ₀ are s _4,6 =214, then record its corresponding coordinate position (u ₀ , v ₀ )=(4,6), where u ₀ =4 can be transformed into A binary bit bin(u ₀ )=(0100) ₂ , can convert v ₀ =6 into A binary bit bin(v ₀ )=(0110) ₂ , thus get For r ₁ =178, if the element in S equal to r ₁ is s _5,2 =178, then will Concatenate bits according to formula (4) to get 2-valued bit string in

Step 3: First start with the texture sample image Select the m+t colors with the highest frequency to form a palette sequence V=(v _i ) _m+t , and then initialize an R-order blank image with a resolution of h ₃ ×w ₃ A random coordinate sequence of length 3 l ₃ is generated by the key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in C are not equal, l ₃ is determined according to formula (5), the symbol Indicates rounding down;

For example: take h ₂ =w ₂ =64, m=16, t=2, then you can select 18 colors with the highest frequency from the texture sample image T=(t _{x, y} ) _64×64 to form a palette Sequence V=(v _i ) ₁₈ ; take h ₃ =w ₃ =512, R=8, an 8th-order blank image K=(ki _{, j} =2 ⁸ -1=255) with a resolution of 512×512 can be obtained _512×512 ; if the random seed of the linear congruential random generator is used as the key k ₂ , k ₂ =7653, l ₂ =16, it can be obtained from formula (5) Then a random coordinate sequence with a length of 3·l ₃ =3·4=12 and two different coordinates can be generated: C=((55,64), (49,74), (82,84), (228 , 113), (89, 73), (169, 196), (170, 165), (253, 129), (104, 222), (116, 62), (235, 246), (235, 161 ));

Step 4: Through the mapping relationship between the elements in the palette sequence V and the element indexes, encode the 3-fold backup of the binary bit string b into a color sequence V′, and finally combine the random coordinate sequence C and place it in the blank image K All elements in the color sequence V'. The specific method is:

1) Bind the 2-valued bits in the 2-valued bit string b to A group of bits is converted into a sequence of length l ₃ where l ₃ is determined according to formula (6), for For each element d _i in , encode d _i as a backup of d _i through formula (7)

2) Execute 1) repeatedly, and D can be coded as 3·l ₃ elements Use it as the color sequence V', as shown in formula (8);

3) Will As the blank image K (x _3·i+k , y _3·i+k ), i=0, 1,..., l ₃ -1, k=0, 1, 2 elements;

E.g:

1) Take b=(0100011001010010) ₂ , l ₂ =16, set the binary bits in b to Bits are converted into a group with a length of The sequence D=(d ₀ , d ₁ , d ₂ , d ₃ )=((0100) ₂ , (0110) ₂ , (0101) ₂ , (0010) ₂ )=(4, 6, 5, 2), For d ₀ =4, it can be calculated by formula (6) Where (x _3·0+0 ,y _3·0+0 )=(x ₀ ,y ₀ )=(55,64) is the 0th coordinate in C, similarly, it can be obtained

2) For d ₁ =6, it can be obtained For d ₂ =5, we can get For d ₃ =2, we can get According to formula (7), we can get:

3) Will As an element at position (x _3·i+k , y _3·i+k )=(x _3·0+0 , y _3·0+0 )=(55,64) in K, the As an element at position (x ₁ , y ₁ )=(49, 74) in K, set As an element at position (x ₂ , y ₂ )=(82, 84) in K; As (x _3·i+k , y _3·i+k )=(x _3·1+0 , y _3·1+0 )=(x ₃ , y ₃ )=(228,113) position in K element, will As an element at position (x ₄ , y ₄ )=(89, 73) in K, the As an element at position (x ₅ , y ₅ )=(169, 196) in K; As (x _3·i+k , y _3·i+k )=(x _3·2+0 , y _3·2+0 )=(x ₆ , y ₆ )=(170,165) position in K element, will As the element at position (x ₇ , y ₇ )=(253, 129) in K, set As an element at position (x ₈ , y ₈ )=(104, 222) in K; As (x _3·i+k , y _3·i+k )=(x _3·3+0 , y _3·3+0 )=(x ₉ , y ₉ )=(116,62) position in K element, will As the element at position (x ₁₀ , y ₁₀ )=(253, 246) in K, the As an element at position (x ₁₁ , y ₁₁ )=(235, 161) in K.

Step 5: First, combine the pixel-by-pixel texture generation strategy to fill the remaining positions in the blank image K with texture. The specific method is:

In a blank image K, for any k _{i, j} and Denote the γ×γ neighborhood centered on _{ki, j} as N(ki _{, j} ), traverse all the γ×γ neighborhoods N′(t _x y) centered on all elements t _{x, y} in the texture sample map _{, y} ), calculate the similarity between N(k _{i, j} ) and N′(t _{x, y} ) according to formula (8), and select the γ×γ neighborhood most similar to N(k _{i, j} ) according to Assign values to k _{i, j} ; execute repeatedly until all elements in the blank image K Complete the assignment, and then output the blank image K as the generated dense texture image.

For example: Take γ×γ=3×3 as an example, if suppose minimum, ie then press Assign values to k _{i and j} , and in the same way, all elements in the blank image K can be completed assignment.

Referring to Fig. 2, the information recovery method of the present invention comprises the following steps:

Step 1: First input the dense texture image received by the channel Generate a random coordinate sequence with a length of 3 l ₃ through the preset key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in C are not equal, in the dense texture map K′, for each ( _xi , y _i )∈C, find (x _{3 · i+k} , y _3·i+k ), i∈[0, l ₃ -1], k=0, 1, all elements in 2 positions Finally get the color sequence V';

For example: if the dense texture map K′=(k _{i, j} ) _512×512 received by the input channel, then h ₃ =w ₃ =512;

Still take the random number seed of the linear congruential random generator as the key k ₂ , then according to k ₂ =7653, l ₃ =4, the random coordinate sequence with a length of 3·l ₃ =3·4=12 can be generated as follows: C=((55, 64), (49, 74), (82, 84), (228, 113), (89, 73), (169, 196), (170, 165), (253, 129) , (104, 222), (116, 62), (235, 246), (235, 161));

In the dense texture map K′, for each (x _i , y _i )∈C, for example, for (x _3·i+k , y _3·i+k )=(x _3·0+0 , y _{3 0+0} )=(55, 64) can find the element For (x ₁ , y ₁ )=(49, 74) one can find For (x ₁₁ , y ₁₁ )=(235, 161) one can find final color sequence

Step 2: Start with texture sample image Select m+t colors to form a palette sequence V=(v _i ) _m+t , and then decode the color sequence V′ into a coordinate sequence b through the mapping relationship between the elements in the palette sequence V and the element indexes , the specific method is:

1) Find out according to the palette sequence V index position of Then calculate according to formula (9)

2) note The number of elements in the interval [0, m-1] is recorded as E _c , and d _i is calculated according to the following five situations:

①If E _c =0, set

②If E _c =1 and The value of the element located in the interval [0, m-1] is v, then set d _i =v;

③If E _c =2 and The element values in the interval [0, m-1] are v ₁ and v ₂ , then set d _i =(v ₁ +v ₂ )/2;

④ If E _c =3 and There are two or more values that are equal, record the equal value as v, then set d _i = v;

⑤If E _c =3 and If the pair is not equal, set

3) Repeat 2) until all the coordinates in the random coordinate sequence C are processed, and then according to the formula (10) to obtain the binary coordinate sequence b with a length of l ₂ :

In formula (10), Left() is a binary bit string interception function, wherein the first parameter is the binary bit string to be intercepted, and the second parameter is the length to be intercepted from the left, l ₂ according to formula ( 11) Determine:

For example: take h ₂ =w ₂ =64, m=16, t=2, then 18 colors with the highest frequency of occurrence can be selected from T=(t _{x, y} ) _64×64 to form a palette sequence V=( v _i ) ₁₈ ;

1) If find out according to V=(v _i ) ₁₈ is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at is indexed at Then according to formula (9), we can get:

empathy

2) Ruoji but The number of elements in the interval [0, 15] is E _c =3, by E _c =3 and There are two or more values that are equal, record the equal value v=4, so d ₀ =4;

3) Ruoji but The number of elements in the interval [0, 15] is E _c =3, by E _c =3 and There are two or more values that are equal, record the equal value v=6, get d ₁ =6; if record but The number of elements in the interval [0, 15] is E _c =3, by E _c =3 and There are two or more values that are equal, record the equal value v=5, get d ₂ =5; if record but The number of elements in the interval [0, 15] is E _c =3, by E _c =3 and There are two or more values that are equal, record the equal value v=2, get d ₃ =2; suppose l ₁ =2, According to formula (10), the length is Binary coordinate sequence b=Left(bin(d ₀ )||bin(d ₁ )||bin(d ₂ )||bin(d ₃ ), l ₂ )=Left(bin(4)||bin( 6)||bin(5)||bin(2), 8)=(0100011001010010) ₂ ;

Step 3: Firstly, the specific method of generating the R-order key map S through the preset key k ₁ is to generate an R-order key map S with a resolution of h ₁ ×w ₁ from the key k ₁ s _{i, j} ∈ [0, 2 ^R -1], and then combined with the binary bit string b to extract the secret information B in the R-order key map S, the specific method is:

1) Bind the binary bit of the binary bit string b to is 1 group, divided into l ₁ groups, recorded as According to formula (12) and formula (13) from each group before interception 2-valued bits and after two 2-value bits as the coordinates (u _i , v _i ), and then assign the element corresponding to the position (u _i , v _i ) in S to r _i as shown in formula (14), where i=0, 1, ..., l ₁ -1;

2) convert r _i , i=0, 1, ..., l ₁ -1 into binary bit strings of length R and concatenate them into binary bit strings of length N B=(B _i ) _N output.

For example: take the random seed of the linear congruential random number generator as the key k ₁ , for example, take k ₁ =6546, R=8, h ₁ =w ₁ =16, then an 8th order with a resolution of 16×16 can be generated Key map S, suppose b=(0 1 0 0 0 1 1 0 0 1 0 ₂ ;

1) Take l ₁ =2, Set the 2-value bits of b to is 1 group, divided into l ₁ =2 groups, respectively Intercept according to formula (11) before 2-valued bits get Intercept according to formula (12) after the 2-valued bits get By the same token Get u ₁ =(0101) ₂ , v ₁ =(0010) ₂ , assuming that elements s _4,6 =214, s _5,2 =178 in S, then (u ₀ , v ₀ )=(4, 6) The element corresponding to the position is assigned to r ₀ , r ₀ = s 4, ₆ = 214, and the element corresponding to the position (u ₁ , v ₁ ) = (5, 2) in S is assigned to r ₁ , it can be obtained r ₁ =s _5,2 =178;

2) Convert r ₀ =214, r ₁ =178 into binary bit strings of length R=8 respectively and concatenate them into binary bit strings of length N=16 B=(bin(r ₀ )||bin(r _i ))=((bin(214)||bin(178)))=(1101011010110010) ₂ output.

Refer to Figures 3 to 31, Figures 8 to 11 use Figures 3 to 6 as texture sample images, and Figure 7 as the encrypted image, and the user keys are respectively key ₁ = 6546, key ₂ = 7653, key ₁ = 6548, key ₂ = 7659, key ₁ = 6546, key ₂ = 7653, key ₁ = 6548, key ₂ = 7659, according to the embedding process corresponding to Figure 1 to obtain the public dense texture image. Fig. 12～Fig. 15 is according to Fig. 2 extracting process, provides correct key, and the cipher map extracted from Fig. 8 ~ Fig. 11 successively, all is 0% with respect to the bit error rate BR of Fig. 7, so cipher map can be Extract correctly. Figures 16 to 19 are images with dense textures after performing 1.9%, 3.4%, 4.1%, and 2.6% clipping attacks on Figures 8 to 11, respectively. Figures 24 to 27 are the dense maps extracted from Figures 16 to 19 according to the extraction process in Figure 2. Compared with Figure 7, the bit error rate BR of Figure 24 is 0.63%, PSNR=30.7281dB, and Figure 25 is compared to Figure 7 The bit error rate BR of 7 is 1.12%, PSNR=28.1291dB, the bit error rate BR of Fig. 26 relative to Fig. 7 is 1.37%, PSNR = 26.7750dB, the bit error rate BR of Fig. 27 relative to Fig. 7 is 0.85%, PSNR=29.8146dB, so the reconstructed secret image is similar enough to the original secret image, indicating that the method of the present invention has a certain ability to resist clipping attacks. Figures 20 to 23 are images with dense textures after 8%, 20%, 15%, and 10% random noise attacks are performed on Figures 8 to 11, respectively. Figures 28 to 31 are the dense maps extracted from Figures 20 to 23 according to the extraction process in Figure 2. Compared with Figure 7, the bit error rate BR of Figure 28 is 2.66%, PSNR=24.3136dB, and Figure 29 is compared to Figure 7 The bit error rate BR of 7 is 6.35%, PSNR=20.1874dB, the bit error rate BR of Fig. 30 relative to Fig. 7 is 4.79%, PSNR = 21.7039dB, and the bit error rate BR of Fig. 31 relative to Fig. 7 is 3.13%, PSNR=23.7018dB, so the reconstructed secret image is similar enough to the original secret image, which shows that the method of the present invention has a certain ability of resisting random noise attack.

If the information camouflage method and the information recovery method of the present invention are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the present invention realizes all or part of the processes in the methods of the above embodiments, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer When the program is executed by the processor, the steps in the above-mentioned various method embodiments can be realized. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form. Computer-readable storage media includes both volatile and non-permanent, removable and non-removable media by any method or technology for storage of information. Information may be computer readable instructions, data structures, modules of a program, or other data. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, computer-readable media Excludes electrical carrier signals and telecommunication signals.

In an exemplary embodiment, a computer-readable storage medium is also provided, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for disguising information or the method for recovering information is implemented. step. Wherein, the computer storage medium may be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disk, hard disk, magnetic tape, magneto-optical disk (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor memory (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state disk (SSD)), etc.

In an exemplary embodiment, there is also provided a computer device, including a memory, a processor, and a computer program stored in the memory and operable on the processor, when the processor executes the computer program Steps for implementing the information forgery method or the information recovery method. The processor may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

The present invention adopts the pixel-by-pixel texture synthesis method to generate dense texture, and the dense texture image is generated point by point instead of splicing small texture blocks, so it avoids the inherent splicing existing in the block splicing generative information hiding based on the suture line algorithm Traces and reduce the generation of repeated patterns, higher security, and less likely to arouse the suspicion of attackers. It effectively solves the problem that in the traditional information hiding method based on block splicing texture synthesis, because the dense texture image uses the seam algorithm to stitch the texture of the overlapping area, it will inevitably produce stitching traces and repeated texture patterns, so that the secret information cannot be hidden. cover up the problem. The secret information is expressed through the coordinates of the key map generated by the key, and the coordinate information of the secret information in the key map is embedded in the secret bunker map instead of the secret information, which avoids the direct channel transmission of the secret information, and the secret information The process of embedding and extracting completely depends on the key, which has high security. It effectively solves the problem that the existing information hiding method based on block splicing texture synthesis is not high in security due to the use of small sample blocks to encode secret information, and the secret information is transmitted in the channel. The present invention introduces an authentication strategy based on multi-backup and interval expansion. When embedding multiple secret information, when extracting, all multiple secret information are extracted, and then compared with multiple secret information, the secret information is the same at most The secret information is taken as the correctly extracted secret information, and if they are not the same, the average value is taken as the secret information. This strategy can effectively improve the anti-attack ability of extracting secret information. It effectively solves the problem that the existing information hiding method based on block splicing texture synthesis lacks authentication information, and it is impossible to determine whether the extracted secret information is correct secret information if the secret bunker map is attacked.

The above content is only to illustrate the technical ideas of the present invention, and cannot limit the protection scope of the present invention. Any changes made on the basis of the technical solutions according to the technical ideas proposed in the present invention shall fall within the scope of the claims of the present invention. within the scope of protection.

Claims (10)

1. A method for disguising information, comprising the following steps: S1: Generate an R-order key map S through the preset key k ₁ , input a binary secret information bit sequence B, and encode the binary secret information bit sequence B into a 2-ary ^R number sequence R; S2: Encode the 2-ary number sequence ^R into a binary bit string b through the pixel coordinates of the same pixel value in the R-order key map S as the element in the 2-ary number sequence ^R ; S3: Select m+t colors from the preset texture sample map T to form a palette sequence V, initialize a blank image K, and generate a random coordinate sequence C through the preset key k ₂ ; S4: Through the mapping relationship between elements and element indexes in the palette sequence V, encode the 3-fold backup of the binary bit string b into a color sequence V', combine the random coordinate sequence C to place the color sequence V' on the blank image K all elements in S5: Fill the remaining positions in the blank image K with texture through a pixel-by-pixel texture generation strategy to obtain a dense texture image K. 2. The method for disguising information according to claim 1, wherein the specific method of the S1 is: S1-1: Generate an R-order key map with a resolution of h ₁ ×w ₁ through the preset key k ₁ S1-2: Input the 2-value secret information bit sequence B, convert each R bit in B into a group of 2 ^R -ary number sequences with length l ₁ Wherein l ₁ is determined by formula (1); Among them, the symbol Indicates rounding up; The specific method of said S2 is: Encode the 2-ary number sequence ^R into a 2-valued bit string through the pixel coordinates of the same pixel value in the R-order key map S as the element in the 2-ary number sequence ^R Wherein l ₂ is determined by formula (2); 3. The method for disguising information according to claim 2, characterized in that the specific method of said S2 is: S2-1: Pair of 2 ^R -ary number sequences For each element r _i in the R-order key map S, randomly select an element equal to the value of r _i and record its corresponding coordinate position (u _i , v _i ), and convert u _i and v _i into lengths of The binary bit string of Among them, bin(u _i ) is a binary conversion function, which is used to convert u _i into a binary bit string. For u _i ∈ [0,h ₁ -1], bin(u _i ) converts u _i into length for 2-valued bit string of , for v _i ∈ [0,w ₁ -1], bin(v _i ) converts v _i to length The 2-value bit string of , the symbol "||" is a bit string concatenation function; S2-2: Repeat S2-1 to Convert to l ₁ binary bit string Through formula (4) will l ₁ binary bit string Concatenate to get 2-valued bit string The specific method of said S3 is: S3-1: Sample images of preset textures at resolution h ₂ ×w ₂ Select the m+t colors with the highest frequency of occurrence to form a palette sequence V=(v _i ) _m+t ; S3-2: Initialize an R-order blank image with a resolution of h ₃ ×w ₃ S3-3: Generate a random coordinate sequence of length 3 l ₃ through the preset key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in the random coordinate sequence C are not equal, l ₃ is determined by formula (5): Among them, the symbol Indicates rounding down; The specific method of said S4 is: S4-1: Bind the 2-valued bits in the 2-valued bit string b to A group of bits is converted into a sequence of length l ₃ S4-2: Convert the sequence Each element d _i in is encoded as a backup of d _i by formula (6) S4-3: Repeat S4-2 to encode D as 3 l ₃ elements Obtain the color sequence V' by formula (7); S4-4: will As the element in the blank image K (x ₃ · _i+k , y ₃ · _i+k ), i=0,1,...,l ₃ -1, k=0,1,2; The concrete method of described S5 is: S5-1: In the blank image K, for any k _i,j and Denote the γ×γ neighborhood centered on _ki,j as N(ki _,j ), traverse all the γ×γ neighborhoods N′(t _x y) centered on all elements t _x,y in the texture sample map T _,y ), get the similarity between N(k _i,j ) and N′(t _x,y ) through formula (8), and select the neighborhood most similar to N(k _i,j ) according to Assign values to k _i , _j ; Among them, p, q are corresponding to 8 neighborhoods N(k _{i, j} ), the corresponding pixel points in N′(t _{x, y} ), R(x), G(x), B(x) respectively represent pixels The R, G, and B component values of point x, the smaller d(N(k _i,j ), N′(t _x,y )), the smaller N(k _i,j ) and N′(t _x,y ) resemblance; S5-2: Repeat S5-1 to all elements in the blank image K After completing the assignment, a dense texture image K is obtained. 4. A computer-readable storage medium, the computer-readable storage medium stores a computer program, characterized in that, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are realized . 5. A computer device, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, characterized in that, when the processor executes the computer program, the computer program according to claim The step of the method described in any one of 1 to 3. 6. A method for recovering information based on the camouflage method described in claim 1, comprising the following steps: R1: Generate a random coordinate sequence C through the preset key k ₂ , and extract the color sequence V' from the input dense texture map K' through the random coordinate sequence C; R2: Select m+t colors from the preset texture sample image T to form a palette sequence V, and decode the color sequence V' into a 2 value through the mapping relationship between the elements in the palette sequence V and the element indexes bit string b; R3: Generate the R-order key map S through the preset key k ₁ , and extract the secret information B from the key map S by combining the binary bit string b. 7. The information recovery method according to claim 6, wherein the specific method of the R1 is: R1-1: Generate a random coordinate sequence with a length of 3 l ₃ through the preset key k ₂ Where ( _xi , y _i )∈h ₃ ×w ₃ and the coordinates in the random coordinate sequence C are not equal to each other; R1-2: From the input dense texture map through the random coordinate sequence C , for each (x _i ,y _i )∈C, find (x _3·i+k ,y _3·i+k ), i∈[0,l ₃ -1],k=0,1,2 all elements of position and combined to obtain the color sequence V'; The concrete method of described R2 is: R2-1: Texture sample image at resolution h ₂ ×w ₂ Select the m+t colors with the highest frequency of occurrence to form a palette sequence V=(v _i ) _m+t ; R2-2: Find out according to the palette sequence V index position of Calculated by formula (9) remember The number of elements in the interval [0,m-1] is recorded as E _c , and d _i is calculated through the following five situations: ①If E _c =0, set ②If E _c =1 and The element value in the interval [0,m-1] is v, set d _i =v; ③If E _c =2 and The element values in the interval [0,m-1] are v ₁ , v ₂ , set d _i =(v ₁ +v ₂ )/2; ④ If E _c =3 and There are two or more values that are equal, record the equal value as v, and set d _i =v; ⑤If E _c =3 and Not equal to each other, set R2-3: Repeat R2-2 until all the coordinates in the random coordinate sequence C have been processed, and the ₂ -value bit string b of length l2 is obtained by formula (10): b＝Left(bin(d ₀ )||bin(d ₁ )||...||bin(d _l3 - ₁ ),l ₂ ) (10) Wherein, Left () is a binary bit string interception function, the first parameter is the binary bit string to be intercepted, and the second parameter is the length intercepted from the left; ₁₂ is determined by formula (11); 8. The information recovery method according to claim 6, wherein the specific method of the R3 is: R3-1: Generate an R-order key map with a resolution of h ₁ ×w ₁ through the preset key k ₁ R3-2: Bind the 2-valued bits of the 2-valued bit string b to is 1 group, divided into l ₁ groups, recorded as From each group through formula (12) and formula (13) respectively before interception 2-valued bits and after two 2-value bits as the coordinates (u _i , v _i ), and then assign the element corresponding to the position (u _i , v _i ) in the R-order key map S to r _i through formula (14), where i=0,1 ,...,l ₁ -1; Wherein, Right () is a binary bit string interception function, the first parameter is the binary bit string to be intercepted, and the length of the second parameter intercepted from the right; R3-3: convert r _i , i=0,1,…,l ₁ -1 into binary bit strings of length R, and concatenate all binary bit strings of length R into length The binary bit string of N, the secret information B=(B _i ) _N is obtained. 9. A computer-readable storage medium, the computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, the steps of the method according to any one of claims 6 to 8 are realized . 10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, characterized in that, when the processor executes the computer program, the computer program according to claim The step of any one of 6 to 8.