CN104424483A - Face image illumination preprocessing method, face image illumination preprocessing device and terminal - Google Patents

Abstract

The invention discloses a lighting preprocessing method for a human face image, comprising: when collecting a human face image, obtaining the lighting value of the environment where the terminal is located; searching for the person closest to the lighting value in a pre-stored human face database face image, and obtain the preprocessing parameters corresponding to the face image with the closest illumination value; perform illumination preprocessing on the collected face image according to the preprocessing parameters. The invention also discloses an illumination preprocessing device and a terminal of a face image at the same time. By adopting the technical scheme of the present invention, high-speed and high-recognition-rate illumination preprocessing can be performed on human face images in an environment with changing illumination.

Description

Method, device and terminal for illumination preprocessing of face image

technical field

The present invention relates to face image processing technology and light sensing technology in face recognition, and in particular to a method, device and terminal for illumination preprocessing of face images. the

Background technique

The impact of illumination changes on face recognition performance is one of the key issues for terminals in a mobile environment. In order to solve the impact of illumination changes on face recognition performance, existing illumination preprocessing algorithms are mainly based on face image transformation and three-dimensional (3D, 3 Dimension) modeling for adaptive illumination compensation. The face image transformation algorithm has poor recognition rate in the mobile environment. Although the 3D modeling algorithm has a high recognition rate, it requires huge computational overhead and time complexity. the

When the terminal is in a mobile environment, the randomness of illumination changes is large and the computing power is limited. If the existing illumination preprocessing algorithm is run on a terminal in a mobile environment, it cannot achieve the expected results in terms of processing speed and recognition rate. the

With the widespread popularity of smart terminals and their continuous impact on people's daily lives, face recognition technology on smart terminals has shown considerable market value and potential. However, the limitation of processing speed and recognition rate restricts the direct application of traditional face recognition technology on smart terminals. Not only that, due to the mobility of smart terminals, face recognition in a mobile environment brings changes in illumination, which increases the difficulty of face recognition applications in a mobile environment. the

Contents of the invention

In view of this, the main purpose of the present invention is to provide a method, device, and terminal for illumination preprocessing of human face images, which can perform illumination preprocessing on human face images at high speed and high recognition rate in an environment with changing illumination. the

In order to achieve the above object, technical solution of the present invention is achieved in that way:

A lighting preprocessing method of a face image, said method comprising:

When collecting face images, and obtain the illumination value of the environment where the terminal is located;

Find the face image closest to the illumination value in the pre-stored face database, and obtain the preprocessing parameters corresponding to the face image closest to the illumination value;

Perform illumination preprocessing on the collected face images according to the preprocessing parameters. the

The method also includes:

Separately collect face images at various light intensities to obtain multiple face images with different light intensities;

Obtain the illumination value of the face images of the plurality of different illumination intensities through the light sensor, and use it as a reference illumination value;

Calculate the preprocessing parameters of the face images of the multiple different illumination intensities respectively according to the illumination preprocessing algorithm;

A face database is established according to the multiple facial images with different illumination intensities, the illumination values of the multiple facial images with different illumination intensities, and the preprocessing parameters of the multiple facial images with different illumination intensities. the

When the illumination preprocessing algorithm is a gamma intensity correction GIC algorithm, the preprocessing parameters are gamma parameters. the

Said searching the face image closest to said illumination value in the pre-stored face database, and obtaining the preprocessing parameters corresponding to the face image closest to said illumination value, including:

Set up the discrete equation between the illumination value of the facial images of the multiple different illumination intensities and the preprocessing parameters of the facial images of the multiple different illumination intensities;

According to the discrete equation, calculate the reference illumination value closest to the illumination value of the collected face image, and obtain the preprocessing parameters of the reference illumination value. the

The method also includes:

Perform face recognition processing on the face image after illumination preprocessing. the

A lighting preprocessing device for a face image, said device comprising: an acquisition unit, an acquisition unit, a search unit and a processing unit; wherein,

The acquisition unit is used to trigger the acquisition unit when acquiring face images;

The acquisition unit is used to acquire the illumination value of the environment where the terminal is located after receiving the trigger of the acquisition unit;

The search unit is used to search the face image closest to the illumination value in the prestored face database, and obtain the preprocessing parameters corresponding to the face image closest to the illumination value;

The processing unit is configured to perform illumination preprocessing on the collected face images according to the preprocessing parameters. the

The device also includes a preprocessing unit and a face database unit;

The collection unit is also used to separately collect face images at various light intensities to obtain multiple face images with different light intensities;

The acquiring unit is also used to acquire the illumination values of the plurality of face images with different illumination intensities, and use them as reference illumination values;

The preprocessing unit is used to calculate the preprocessing parameters of the plurality of face images with different illumination intensities according to the illumination preprocessing algorithm;

The human face database unit is configured to, according to the multiple facial images with different illumination intensities, the illumination values of the multiple facial images with different illumination intensities, and the values of the multiple facial images with different illumination intensities Preprocessing parameters to build a face database. the

When the illumination preprocessing algorithm is a GIC algorithm, the preprocessing parameters are gamma parameters. the

The search unit includes: establishing a subunit and calculating a subunit; wherein,

The establishment subunit is used to establish a discrete equation between the illumination values of the multiple facial images with different illumination intensities and the preprocessing parameters of the multiple facial images with different illumination intensities;

The calculation subunit is used to calculate the reference illumination value closest to the illumination value of the collected face image according to the discrete equation, and obtain the preprocessing parameters of the reference illumination value. the

The device also includes a recognition unit, configured to perform face recognition processing on the face image after illumination preprocessing. the

A terminal, which includes the above-mentioned illumination preprocessing device for any face image. the

The illumination preprocessing method, device and terminal of the human face image recorded in the embodiment of the present invention, when collecting the human face image, and obtain the illumination value of the environment where the terminal is located; search the closest to the illumination value in the pre-stored human face database face image, and obtain the preprocessing parameters corresponding to the face image with the closest illumination value; perform illumination preprocessing on the collected face image according to the preprocessing parameters. In this way, high-speed and high-recognition-rate illumination preprocessing can be performed on face images in an environment with changing illumination, thereby improving processing speed and reducing resource overhead while ensuring a high recognition rate. the

Description of drawings

Fig. 1 is the implementation flow schematic diagram of the illumination preprocessing method of face image of the embodiment of the present invention;

Fig. 2 is the structural composition schematic diagram of the lighting preprocessing device of face image of the embodiment of the present invention;

3 is a schematic diagram of the structural composition of the search unit in the illumination preprocessing device of the face image of the embodiment of the present invention;

FIG. 4 is a schematic diagram of the structural composition of the terminal according to the embodiment of the present invention;

Fig. 5 is the face image schematic diagram under different illumination conditions in the embodiment of the present invention;

FIG. 6 is a schematic diagram of a face image of a sample under different lighting conditions in an embodiment of the present invention. the

Detailed ways

In order to understand the characteristics and technical contents of the present invention in more detail, the implementation of the present invention will be described in detail below in conjunction with the accompanying drawings. The attached drawings are only for reference and description, and are not intended to limit the present invention. the

The basic idea of the embodiment of the present invention is to use the light sensor embedded on the mobile phone panel to collect the data of light intensity as a priori parameter to perform light preprocessing of the face image, so as to achieve the goal in a mobile environment with limited processing speed and recognition rate. In the terminal, high-speed and high-recognition rate illumination preprocessing is performed on face images. the

The embodiment of the present invention records a method of illumination preprocessing of a human face image, as shown in Figure 1, the method includes the following steps:

Step 101: When collecting a face image, obtain the illumination value of the environment where the terminal is located. the

Preferably, the method also includes:

Collect face images at various light intensities respectively to obtain multiple face images with different light intensities;

Obtain the illumination value of the face images of the plurality of different illumination intensities through the light sensor, and use it as a reference illumination value;

Calculate the preprocessing parameters of the face images of the multiple different illumination intensities respectively according to the illumination preprocessing algorithm;

A face database is established according to the multiple facial images with different illumination intensities, the illumination values of the multiple facial images with different illumination intensities, and the preprocessing parameters of the multiple facial images with different illumination intensities. the

Preferably, when the illumination preprocessing algorithm is a Gamma Intensity Correction (GIC, Gamma Intensity Correction) algorithm, the preprocessing parameter is a gamma parameter. the

In order to understand the embodiment of the present invention more clearly, the GIC algorithm is now described in detail. the

The GIC algorithm makes the face image match the predefined standard face image by modifying the global brightness of the face image. A standard face image is usually a face image collected or corrected under standard lighting conditions. the

For example, I is the input face image, and I' is the face image processed by I through the GIC algorithm, then I' can be obtained by formula (1). the

I' _xy = G(I _xy ;γ ^* ) (1)

Among them, I _xy represents the gray value at (x, y) of the input face image coordinates; I' _xy represents the gray value at (x, y) of the face image obtained through G() transformation ; G(I _xy ; γ ^* ) means performing gamma transformation on I _xy .

The gamma parameter γ ^* in formula (1) can be calculated by formula (2). Formula (2) is to select the optimal gamma parameter γ ^* among multiple gamma parameters γ. The purpose of formula (2) is to find a gamma parameter γ ^* that has the smallest difference with the standard face image.

${\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

in, $\arg \underset{\mathrm{\&gamma;}}{\min }{\underset{x,\mathrm{the\; y}}{\mathrm{\&Sigma;}}[G(I_{\mathrm{xy}};\mathrm{\&gamma;})-I_0(x,\mathrm{the\; y})]}^2$ express $\underset{x,\mathrm{the\; y}}{\mathrm{\&Sigma;}}{[G(I_{\mathrm{xy}};\mathrm{\&gamma;})-I_0(x,\mathrm{the\; y})]}^2$ The γ value when taking the minimum value; the expression of G(I _xy ; γ) is formula (3).

$\mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; \&CenterDot;</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}^{\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Among them, c is the grayscale scaling parameter. the

After the above process, multiple face images with different light intensities can obtain corresponding preprocessing parameters, that is, the gamma parameter γ ^* .

In the above GIC algorithm, the biggest contribution to the time complexity is mainly the following two steps: 1) the optimization process of obtaining the gamma parameter γ ^* ; 2) the process of generating the final face image with the gamma parameter γ ^* . And, in addition to the above two processes, there are two other parameters that have an important contribution to the time complexity, namely the range of the gamma parameter γ and the step size μ for searching the gamma parameter γ ^* . Assuming the gamma parameter γ ^* ranges from Equation (4),

γ ^* ∈ [R ₀ , R ₁ ], γ ^* >0 (4)

Wherein, γ ^* ∈ [R ₀ , R ₁ ] indicates that the value range of the solo gamma parameter γ ^* is greater than/equal to R ₀ and less than/equal to R ₁ .

Assume that the time complexity of the optimization process for obtaining the gamma parameter γ ^* is T ₁ , and the time complexity of the process of using the gamma parameter γ ^* to generate the final face image is T ₂ . Then the total time complexity T can be obtained from the following formula (5) to formula (9). In formula (5) to formula (9), O() is the time complexity function, and, in the face image database Contains n face images, each face image has the same size, H is the height of the face image, W is the width of the face image. The calculation process of the total time complexity T is described in detail below.

Combined with formula (4), the time complexity of formula (2) can be expressed as formula (5)

${\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; |</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; W</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&Psi;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; (</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; |</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Among them, the expression of Ψ is as formula (6). the

Ψ=[G(I _xy ;γ)-I ₀ (x,y)] ² (6)

The simplified form of T ₁ can be obtained by formula (7).

${\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; (</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; |</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; W</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$

Formula (1) uses the calculated solo gamma parameter γ ^* to generate the final face image, and its time complexity is calculated as shown in formula (8).

T ₂ =O(H·W) (8)

To sum up, the calculation method of the total time complexity of the GIC algorithm is shown in formula (9). the

$\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; (</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; |</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; W</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span>$

Step 102: Search for the face image closest to the illumination value in the prestored face database, and obtain the preprocessing parameters corresponding to the face image with the closest illumination value. the

Preferably, the search for the face image closest to the illumination value in the pre-stored face database, and obtain the preprocessing parameters corresponding to the face image closest to the illumination value, including:

Set up the discrete equation between the illumination value of the facial images of the multiple different illumination intensities and the preprocessing parameters of the facial images of the multiple different illumination intensities;

According to the discrete equation, calculate the reference illumination value closest to the illumination value of the collected face image, and obtain the preprocessing parameters of the reference illumination value. the

Specifically, assuming that L is the set of illumination values of all face images in the face image database, a one-to-one corresponding discrete equation F can be created, such as formula (10). the

γ _i =F(l _i ),i=1,...,|L| (10)

Among them, l _i is the illumination value of the i-th face image in L, and γ _i is the pre-calculated gamma parameter γ of the i-th face image. Assuming that the illumination value of the collected face image is l, l ^* can be calculated through the process of calculating the gamma parameter γ ^* , as shown in formula (11).

${\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{{\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

In this way, through the formula (11), the time complexity of the gamma parameter γ ^* optimization process becomes the time complexity of the illumination value l ^* optimization process, as shown in the formula (12).

T ₁ '=n·O((l _i -l) ² )+O(n)+O(1)=O(n) (12)

To sum up, the overall time complexity of the illumination preprocessing method of the face image in the embodiment of the present invention can be obtained by formula (13). the

T'=T ₁ '+T ₂ =O(n)+O(H·W) (13)

It can be seen from formula (13) that the overall time complexity is a fixed value, which only depends on the size of the face image database and the size of each face image. Compared with the previous GIC algorithm, the illumination preprocessing method of the face image in the embodiment of the present invention can greatly reduce the time complexity and the resources consumed by calculation. the

Step 103: Perform illumination preprocessing on the collected face images according to the preprocessing parameters. the

Preferably, the method further includes: performing face recognition processing on the face image after illumination preprocessing. the

The embodiment of the present invention also records a lighting preprocessing device for a face image, as shown in FIG. 2 , the device includes: an acquisition unit 21, an acquisition unit 22, a search unit 23, and a processing unit 24; wherein,

The acquisition unit 21 is used to trigger the acquisition unit 22 when acquiring a face image;

The acquisition unit 22 is configured to acquire the illumination value of the environment where the terminal is located after receiving the trigger of the acquisition unit 21;

The search unit 23 is used to search the face image closest to the illumination value in the prestored face database, and obtain the preprocessing parameters corresponding to the face image closest to the illumination value;

The processing unit 24 is configured to perform illumination preprocessing on the collected face images according to the preprocessing parameters. the

Preferably, on the basis of the lighting preprocessing device for human face images shown in Figure 2, the lighting preprocessing device for human face images in the embodiment of the present invention further includes: a preprocessing unit (not shown in Figure 2) and human Face database unit (not shown in Fig. 2); Wherein,

The collection unit 21 is also used to separately collect face images of various light intensities to obtain multiple face images with different light intensities;

The acquisition unit 22 is also used to acquire the illumination values of the multiple face images with different illumination intensities, and use them as reference illumination values;

The preprocessing unit is used to calculate the preprocessing parameters of the plurality of face images with different illumination intensities according to the illumination preprocessing algorithm;

The human face database unit is configured to, according to the multiple facial images with different illumination intensities, the illumination values of the multiple facial images with different illumination intensities, and the values of the multiple facial images with different illumination intensities Preprocessing parameters to build a face database. the

Preferably, when the illumination preprocessing algorithm is a GIC algorithm, the preprocessing parameters are gamma parameters. the

Preferably, as shown in FIG. 3, the search unit 23 includes: an establishment subunit 231 and a calculation subunit 232; wherein,

The establishment subunit 231 is used to establish a discrete equation between the illumination values of the multiple facial images with different illumination intensities and the preprocessing parameters of the multiple facial images with different illumination intensities;

The calculation subunit 232 is configured to calculate the reference illumination value closest to the illumination value of the collected face image according to the discrete equation, and obtain preprocessing parameters of the reference illumination value. the

Preferably, on the basis of the lighting preprocessing device for human face images shown in FIG. 2 , the lighting preprocessing device for human face images in the embodiment of the present invention further includes a recognition unit (not shown in FIG. 2 ) for The face image after illumination preprocessing is processed for face recognition. the

Those skilled in the art should understand that the realization functions of each unit and its subunits in the illumination preprocessing device of the face image shown in Figure 2 and Figure 3 can be understood with reference to the relevant description of the illumination preprocessing method of the aforementioned face image . The functions of each unit and its subunits in the illumination preprocessing device for human face images shown in Fig. 2 and Fig. 3 can be realized by a program running on a processor, or by a specific logic circuit. the

In practical applications, the acquisition unit 21 in the illumination preprocessing device of the above-mentioned face image can be realized by a camera; the acquisition unit 22 can be realized by a built-in light sensor in the terminal; Processing Unit), or digital signal processor (DSP, Digital Signal Processor), or programmable logic array (FPGA, Field-Programmable Gate Array) implementation. the

The embodiment of the present invention also records a terminal. As shown in FIG. 4, the terminal includes the illumination preprocessing device for the face image shown in FIG. Unit 24; where,

The collection unit 21 is used to trigger the acquisition unit when collecting face images;

The acquisition unit 22 is configured to acquire the illumination value of the environment where the terminal is located after receiving the trigger from the acquisition unit;

The search unit 23 is used to search the face image closest to the illumination value in the prestored face database, and obtain the preprocessing parameters corresponding to the face image closest to the illumination value;

The processing unit 24 is configured to perform illumination preprocessing on the collected face images according to the preprocessing parameters. the

Preferably, on the basis of the lighting preprocessing device for human face images shown in Figure 2, the lighting preprocessing device for human face images in the embodiment of the present invention further includes: a preprocessing unit (not shown in Figure 2) and human Face database unit (not shown in Fig. 2); Wherein,

The collection unit 21 is also used to separately collect face images of various light intensities to obtain multiple face images with different light intensities;

The acquisition unit 22 is also used to acquire the illumination values of the multiple face images with different illumination intensities, and use them as reference illumination values;

The preprocessing unit is used to calculate the preprocessing parameters of the plurality of face images with different illumination intensities according to the illumination preprocessing algorithm;

The human face database unit is configured to, according to the multiple facial images with different illumination intensities, the illumination values of the multiple facial images with different illumination intensities, and the values of the multiple facial images with different illumination intensities Preprocessing parameters to build a face database. the

Preferably, when the illumination preprocessing algorithm is a GIC algorithm, the preprocessing parameters are gamma parameters. the

Preferably, as shown in FIG. 3, the search unit 23 includes: an establishment subunit 231 and a calculation subunit 232; wherein,

The establishment subunit 231 is used to establish a discrete equation between the illumination values of the multiple facial images with different illumination intensities and the preprocessing parameters of the multiple facial images with different illumination intensities;

The calculation subunit 232 is configured to calculate the reference illumination value closest to the illumination value of the collected face image according to the discrete equation, and obtain preprocessing parameters of the reference illumination value. the

Preferably, on the basis of the lighting preprocessing device for human face images shown in FIG. 2 , the lighting preprocessing device for human face images in the embodiment of the present invention further includes a recognition unit (not shown in FIG. 2 ) for The face image after illumination preprocessing is processed for face recognition. the

Those skilled in the art should understand that the illumination preprocessing device of the human face image in the terminal shown in Figure 4 can refer to each unit and its subunits in the illumination preprocessing device of the human face image shown in Figure 2 and Figure 3 function to understand. the

In practical applications, the acquisition unit 21 in the lighting preprocessing device of the face image in the terminal shown in FIG. 4 can be realized by a camera; the acquisition unit 22 can be realized by a light sensor built into the terminal; Central processing unit (CPU, Central Processing Unit), or digital signal processor (DSP, Digital Signal Processor), or programmable logic array (FPGA, Field-Programmable Gate Array) implementation. the

The effect of the illumination preprocessing method for a face image provided by the present invention will be further described in detail below in conjunction with specific embodiments. the

In this embodiment, the relevant face image data in the HUN-MFD standard mobile face database is selected. There are 349 face images and 12 samples, including 4 females and 8 males. The number of face images in each sample ranges from 14 to 55, including face images in multiple scenes. The face images of each sample are selected from frontal poses, under different lighting conditions, without makeup and expressions. The training set, which is the face image database in the embodiment of the present invention, contains 246 face images, and the remaining 103 face images are used as a test set. There are no duplicate face images between the training set and the test set. Each photo was cut and normalized to a grayscale face image in advance, and a face image of 128×128 pixels was uniformly used. Figure 5 shows the face images of three samples under different lighting conditions. the

In the embodiment, three classic face features are used as comparison to evaluate the performance: PCA[11], LDA[12] and LBP[13]. The nearest neighbor (NN) method is used to calculate the similarity. All face recognition rates here refer to Rank-1. The parameters of the three face features are described as follows:

PCA: A principal component with a face feature length of 50 is used, and each feature saves 92% of the information in the original image;

LDA: Since there are 13 samples, the default value is used, so the feature length is 12;

LBP: As a local feature, the radius of the LBP feature is set to 1, and 8 pixels around each pixel will be used to participate in the calculation. By default, the face image is divided into 8×8 blocks. the

In this performance test embodiment, two operating environments are established for comparison, namely the computer environment and the mobile environment. The embodiment in the computer environment is mainly to verify whether the mobile GIC algorithm can achieve the expected effect in both computer and mobile environments, and to compare which environment is more suitable. The embodiments in both environments set the priority of the program to the highest priority to prevent other programs running in parallel from affecting the embodiment to the greatest extent. The specific parameters of the environment of the two embodiments are as follows:

Computer operating environment: the application program of the embodiment of the present invention runs on the Windows personal computer, and its CPU is Dual-Core E52002.50GHz, memory 2G, operating system is Windows 7 Ultimate Service Pack 1. The dependent libraries used in program development include library functions in Visual Studio 2010 and OpenCV 2.4.2;

Mobile operating environment: the embodiment program runs on an Android mobile phone, the model is Samsungi9250 (Google Galaxy Nexus), the CPU is OMAP4460 Dual-Core 1228 MHz, the memory is 1G, the operating system is Android 4.1.2 (Jelly Bean), the operating system The kernel version is Linux3.0.31-g4f6d371. The dependent libraries used in program development include android-ndk-r8b and OpenCV2.4.2. In order to maintain consistency with programs in the computer environment, the same C++ code is transplanted to the mobile environment without additionally writing a program rewritten from C++ to Java version. Therefore, in order to meet such requirements, the application program is developed based on Java Native Interface (JNI, Java Native Interface) technology. the

When running the GIC algorithm, the range of γ is usually set to (0, 5] according to experience. Figure 6 shows the original face image, GIC and the face image after moving GIC under different lighting conditions for a sample.

Table 1 shows the speedup ratio of the preprocessing part when the mobile GIC algorithm (which is the illumination preprocessing method of the face image in the embodiment of the present invention) is compared with the GIC algorithm and uses different step lengths in the computer environment and the mobile environment . the

Table 1

Table 2 shows the Rank-1 face recognition rate on mobile phones using PCA, LDA and LBP face features and different preprocessing methods. the

Table 2

Table 3 shows the speedup ratio between the mobile gamma intensity correction and the gamma intensity correction of the complete face recognition process including the preprocessing part in the mobile environment. the

table 3

From the statistical data in Table 1 to Table 3 above, it can be concluded that:

The acceleration effect of the mobile GIC algorithm is better in the mobile environment than in the computer environment;

Compared with no compensation, the mobile GIC algorithm greatly improves the face recognition rate, but compared with the GIC algorithm, in the case of PCA and LDA algorithms, the recognition rate of the mobile GIC algorithm is slightly lower, but with the local feature algorithm LBP , the mobile GIC algorithm has a higher recognition rate;

Due to the extremely low computational overhead and power consumption in the mobile environment, the mobile GIC algorithm is particularly suitable for running on mobile devices, especially when using LDA, which not only improves the recognition rate, but also greatly reduces the time complexity. the

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. the

Claims (11)

1. A lighting preprocessing method for a face image is characterized by comprising the following steps:

when a face image is collected, acquiring an illumination value of the environment where the terminal is located;

searching a face image closest to the illumination value in a prestored face database, and acquiring a preprocessing parameter corresponding to the face image closest to the illumination value;

and carrying out illumination preprocessing on the acquired face image according to the preprocessing parameters.

2. The method of claim 1, further comprising:

respectively collecting face images with various illumination intensities to obtain a plurality of face images with different illumination intensities;

acquiring illumination values of the plurality of face images with different illumination intensities through the light sensor, and taking the illumination values as reference illumination values;

respectively calculating preprocessing parameters of the plurality of face images with different illumination intensities according to an illumination preprocessing algorithm;

and establishing a face database according to the plurality of face images with different illumination intensities, the illumination values of the plurality of face images with different illumination intensities and the preprocessing parameters of the plurality of face images with different illumination intensities.

3. The method of claim 1, wherein when the illumination pre-processing algorithm is a Gamma Intensity Correction (GIC) algorithm, the pre-processing parameters are gamma parameters.

4. The method according to any one of claims 1 to 3, wherein the searching for the face image closest to the illumination value in a pre-stored face database and obtaining the preprocessing parameter corresponding to the face image closest to the illumination value comprises:

establishing a discrete equation between the illumination values of the plurality of face images with different illumination intensities and the preprocessing parameters of the plurality of face images with different illumination intensities;

and calculating a reference illumination value closest to the illumination value of the acquired face image according to the discrete equation, and acquiring a preprocessing parameter of the reference illumination value.

5. The method according to any one of claims 1 to 3, further comprising:

and carrying out face recognition processing on the face image subjected to illumination preprocessing.

6. An apparatus for preprocessing illumination of a face image, the apparatus comprising: the device comprises a collecting unit, an acquiring unit, a searching unit and a processing unit; wherein,

the acquisition unit is used for triggering the acquisition unit when acquiring the face image;

the acquisition unit is used for acquiring the illumination value of the environment where the terminal is located after receiving the trigger of the acquisition unit;

the searching unit is used for searching the face image closest to the illumination value in a prestored face database and acquiring the preprocessing parameter corresponding to the face image closest to the illumination value;

and the processing unit is used for carrying out illumination preprocessing on the acquired face image according to the preprocessing parameters.

7. The apparatus of claim 6, further comprising a preprocessing unit and a face database unit;

the acquisition unit is also used for respectively acquiring the face images under various illumination intensities to obtain a plurality of face images with different illumination intensities;

the acquiring unit is further configured to acquire illumination values of the plurality of face images with different illumination intensities, and use the illumination values as reference illumination values;

the preprocessing unit is used for respectively calculating preprocessing parameters of the plurality of face images with different illumination intensities according to an illumination preprocessing algorithm;

the face database unit is used for establishing a face database according to the plurality of face images with different illumination intensities, the illumination values of the plurality of face images with different illumination intensities and the preprocessing parameters of the plurality of face images with different illumination intensities.

8. The apparatus of claim 6, wherein when the illumination pre-processing algorithm is a GIC algorithm, the pre-processing parameters are gamma parameters.

9. The apparatus according to any one of claims 6 to 8, wherein the search unit comprises: establishing a subunit and a calculating subunit; wherein,

the establishing subunit is configured to establish a discrete equation between the illumination values of the multiple face images with different illumination intensities and the preprocessing parameters of the multiple face images with different illumination intensities;

and the calculating subunit is used for calculating a reference illumination value closest to the illumination value of the acquired face image according to the discrete equation and acquiring a preprocessing parameter of the reference illumination value.

10. The apparatus according to any one of claims 6 to 8, further comprising a recognition unit for performing face recognition processing on the illumination-preprocessed face image.

11. A terminal characterized in that it comprises means for the pre-processing of the illumination of face images according to any one of claims 6 to 10.