WO2015109693A1

WO2015109693A1 - Method and system for image color calibration

Info

Publication number: WO2015109693A1
Application number: PCT/CN2014/077684
Authority: WO
Inventors: 马亮
Original assignee: 中兴通讯股份有限公司
Priority date: 2014-01-22
Filing date: 2014-05-16
Publication date: 2015-07-30
Also published as: CN104796683A; CN104796683B

Abstract

A method for image color calibration comprises: performing human face recognition on an image undergoing automatic white balancing, and if the image is successfully recognized, determining a human face area, and calculating the statistical values of red-green-blue RGB of the human face area; calculating corrected white balance gains for red, green and blue RGB three channels required for correcting the statistical values of red-green-blue RGB of the human face area as RGB values of an ideal human face model; and according to the corrected white balance gains for red, green and blue RGB three channels, performing once again white balancing on the image undergoing automatic white balancing. Also disclosed in the embodiment of the present invention is a system for image color calibration.

Description

Method and system for calibrating image color

Technical field

The present invention relates to the field of image processing technologies, and in particular, to a method and system for calibrating image colors.

Background technique

The color temperature (Colo(u)r Temperature) is a measure of the color of the light source in K (Kelvin). Color temperature has important applications in photography, video, and publishing. The color temperature of the source is determined by comparing its color to the theoretical thermal blackbody radiator. The Kelvin temperature of a hot black body radiator matching the color of the light source is the color temperature of that source, which is directly related to Planck's law of blackbody radiation.

The human eye is identified as white for the brightest object at any color temperature. The photos taken by the camera at different color temperatures show different colors, such as the blue light under the D65 light source and the yellowish yellow light. The indoor light source is often complicated, and the color temperature of incandescent lamps and fluorescent lamps is not very standard. Therefore, shooting a portrait indoors often results in an abnormal color tone, yellowish or bluish.

The essence of White Balance (WB) is to make white objects appear white under any color source. This is easy for the human eye because the human eye has the ability to adapt, but the camera is different. The white object captured by the camera will carry the color of the light source. Automatic white balance (AWB) is done by color correction to make the color of the captured image the normal color seen by the human eye. The photo read from the sensor chip is called the original picture, and the automatic white balance color correction of the original picture is obtained by multiplying the corresponding gains Gr, Gg, Gb on the three channels of the red, green and blue RGB of the original picture to achieve white balance. effect.

For mixed light sources, the effect of automatic white balance is usually different from the normal color seen by the human eye. Summary of the invention

The present invention provides a method and system for calibrating image colors that can correct the gain of automatic white balance based on facial skin color to improve image quality. The present invention provides a method of calibrating image colors, including:

Face recognition is performed on the image processed by the automatic white balance. If the recognition is successful, the face area is determined, and the red, green and blue RGB statistical values of the face area are calculated;

Calculating the corrected white balance gain of the three channels of red, green, and blue RGB required to correct the RGB statistics of the face region to the RGB values of the ideal face model;

The image processed by the automatic white balance is white-balanced according to the corrected white balance gain of the three channels of red, green and blue RGB.

Optionally, the method further includes the following features:

Calculating red, green and blue RGB statistics of the face region, including:

The red, green and blue RGB values of all the pixels of the face region are summed and averaged to obtain a red, green and blue RGB average value of the face region.

Optionally, the method further includes the following features:

The corrected white balance gain of the three channels of red, green and blue RGB required to correct the RGB statistics of the face region to the RGB values of the ideal face model, including:

Calculating a brightness value of the face region according to an RGB average value of the face region;

Calculating the brightness value of the ideal face model based on the RGB values of the ideal face model;

Comparing the brightness value of the face area with the brightness value of the ideal face model, respectively calculating a gain correction factor for each of the three channels of red, green and blue RGB according to the brightness comparison result, and each channel of red, green and blue RGB The current white balance gain is multiplied by the gain correction factor of the channel to obtain a corrected white balance gain for the channel.

Optionally, the method further includes the following features:

The calculating a gain correction factor for each of the three channels of red, green, and blue RGB according to the brightness comparison result, including:

When the brightness value of the face area. When the brightness value of the ideal face model is greater than or equal to the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region, the luminance value of the ideal face model is divided by the face region. The quotient obtained from the luminance value is used as an index, and the power of the base and the exponential is used as a gain correction factor of the channel corresponding to the primary color; The gain correction factor " _R , the gain correction factor of the green channel" _σ , the gain correction factor <3⁄4 of the blue channel are as follows:

3⁄4 = (W ^) ;

a _a = . strict);

= The brightness value of the face area. When the luminance value of the ideal face model is smaller than the luminance of the ideal face model, the difference between the luminance maximum value and the luminance value of the ideal face model is divided by the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region. a quotient obtained by the difference between the luminance maximum value and the luminance value of the face region as an index, the power of the base and the exponential constructed as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel " _R , The gain correction factor o _{Q of the} green channel and the gain correction factor of the blue channel are as follows:

a _R = strict leg u;

Where, the red component value of the face region, R, is the red component value of the ideal face model,

G. The green component value of the face region is the green component value of the ideal face model, S. It is the blue component value of the face region, A is the blue component value of the ideal face model, Y _max is the brightness maximum value, and the Y dish is preset.

Optionally, the method further includes the following features:

The face area is a rectangular area or a circular area containing a face image. The present invention also provides a system for calibrating image colors, including:

a face recognition module configured to perform face recognition on an image subjected to automatic white balance processing, and if the recognition is successful, determine a face region, calculate a red, green and blue RGB statistical value of the face region; a gain calculation module, Set to calculate a corrected white balance gain of three channels of red, green, and blue RGB required to correct the RGB statistical value of the face region to the RGB value of the ideal face model; and a white balance processing module configured to be red and green The corrected white balance gain of the three channels of blue RGB re-white balances the image subjected to the automatic white balance processing. Optionally, the system further includes the following features:

The face recognition module calculates the red, green and blue RGB statistics of the face region by: summing the red, green and blue RGB values of all the pixels of the face region and averaging the average to obtain the face region Red, green and blue RGB average.

Optionally, the system further includes the following features:

The gain calculation module calculates the corrected white balance gain of the three channels of red, green and blue RGB required to correct the RGB statistics of the face region to the RGB values of the ideal face model by: RGB average according to the face region Calculating a brightness value of the face region;

Optionally, the system further includes the following features:

The gain calculation module calculates the gain correction factors of each of the three channels of red, green, and blue RGB according to the brightness comparison result as follows:

When the brightness value of the face area. When the brightness value of the ideal face model is greater than or equal to the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region, the luminance value of the ideal face model is divided by the face region. The quotient obtained from the luminance value is used as an index, and the power of the base and the exponential is used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel, _R , the gain correction factor of the green channel, _σ , blue The gain correction factors for the channels are as follows:

3⁄4 = (W ^) ;

a _a = . strict);

=„ w _;

When the brightness value of the face area. When the brightness value of the ideal face model is smaller than the chromatic value of the corresponding basic color component of the face region, the radiance of the ideal face model is used as the base, and the brightness is the most The difference between the difference between the large value and the brightness value of the ideal face model divided by the difference between the brightness maximum value and the brightness value of the face region is used as an index, and the base and the exponential constructed power are used as the channel corresponding to the primary color. Gain correction factor; the gain correction factor of the red channel " _R , the gain correction factor of the green channel" _σ , the gain correction factor of the blue channel are as follows:

a _R = strict leg u;

Where is the red component value of the face region, R, which is the red component value of the ideal face model, G. The green component value of the face region is the green component value of the ideal face model, S. It is the blue component value of the face region, A is the blue component value of the ideal face model, Y _max is the brightness maximum value, and the Y dish is preset.

Optionally, the system further includes the following features:

The face area is a rectangular area or a circular area containing a face image.

Compared with the related art, a method and a system for calibrating an image color according to an embodiment of the present invention corrects an RGB statistical value of a face region to an ideal human face by performing face recognition on an image subjected to automatic white balance processing. The corrected white balance gain of the three channels of red, green and blue RGB required for the RGB value of the model, the image is re-whitened according to the corrected white balance gain, and the embodiment of the present invention can render the processed image The color is closer to the color seen by the human eye, improving the image quality. BRIEF abstract

1 is a flow chart of a method for calibrating an image color according to an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of a system for calibrating image colors according to an embodiment of the present invention.

Preferred embodiment of the invention

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other. As shown in FIG. 1 , an embodiment of the present invention provides a method for calibrating an image color, the method comprising:

S10, performing face recognition on the image processed by the automatic white balance. If the recognition is successful, determining a face region, and calculating a red, green, and blue RGB statistical value of the face region;

S20, calculating a corrected white balance gain of three channels of red, green, and blue RGB required to correct the RGB statistical value of the face region to the RGB value of the ideal face model;

S30: Perform white balance processing on the image subjected to the automatic white balance processing according to the corrected white balance gain of the three channels of red, green and blue RGB.

The method also includes the following features:

The face area is a rectangular area or a circular area including a face image;

The calculating the red, green, and blue RGB statistics of the face region includes: summing the red, green, and blue RGB values of all the pixels of the face region, and averaging the red, green, and blue colors of the face region. RGB average value;

The corrected white balance gain of the three channels of red, green, and blue RGB required to correct the RGB statistics of the face region to the RGB values of the ideal face model includes:

(a) Calculating the luminance value Y of the face region based on the RGB average of the face region. ;

(b) calculating the brightness value X of the ideal face model according to the RGB value of the ideal face model;

(c) comparing the brightness value of the face region with the brightness value of the ideal face model, and calculating the corrected white balance gain of the three channels of red, green, and blue RGB according to the brightness comparison result;

Among them, a common algorithm for calculating the luminance value Y from the red, green and blue RGB values is as shown in the formula (1): Y=k _r *R + (l_k _r _k _b )*G + k _b *B; (1) , =0.299, =0.114 ; R represents the red component value, G represents the green component value, and B represents the blue component value;

Therefore, in step (a), step (b):

Y ₀ =k _r *R ₀ +(lk _r -k _b )*G ₀ +k _b *B ₀ (1-1)

Y,=k _r *R,+(lk _r -k _b )*G,+k _b *Β _λ ( 1-1 ) where is the red component value of the face region, which is the red component value of the ideal face model, G ₀ The green component value of the face region is the green component value of the ideal face model, S. The blue component value of the face region is the blue component value of the ideal face model.

In step (c), comparing the brightness value of the face area with the brightness value of the ideal face model, and calculating the gain correction factor of each of the three channels of red, green and blue RGB according to the brightness comparison result, which will be red The current white balance gain of each channel of the green-blue RGB is multiplied by the gain correction factor of the channel to obtain the corrected white balance gain of the channel;

Among them, the brightness value of the face area. When the brightness value of the ideal face model is greater than or equal to the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region, the luminance value of the ideal face model is divided by the face region. The quotient obtained from the luminance value is used as an index, and the power of the base and the exponential is used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel, _R , the gain correction factor of the green channel, _σ , blue The gain correction factors for the channels are as follows:

3⁄4 = (W ^) ;

a _a = . strict);

=„ w _;

When the brightness value of the face area. When the luminance value of the ideal face model is smaller than the luminance of the ideal face model, the difference between the luminance maximum value and the luminance value of the ideal face model is divided by the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region. a quotient obtained by the difference between the maximum brightness value and the brightness value of the face region as an index, the power of the base and the exponential constructed as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel " _R , green The gain correction factor of the channel and the gain correction factor of the blue channel are as follows:

a _R = strict leg u;

Where Y _max is the maximum brightness value, and Y _{max is} generally preset to 255. The following is how to calculate the corrected white balance gain of the three channels of red, green and blue RGB according to the RGB values of the ideal face model, two examples: Example 1, assuming that the image after the automatic white balance processing is blue overall, the current RGB gains are: ^G «= 3.48, ^G = l, ^G »= 2.07, and the RGB values of the ideal face model are: = 172.5, ^G i= 117.3, ^B , = 99.3; It is found through simulation that if ^G « is corrected to 3.8, ^{G is} corrected to 1, and ^G « is corrected to 1.8, the color of the image is close to the normal color seen by the human eye. Therefore, it can be seen that In the white balance processed image, the blue channel gain ^G «higher, the red channel gain ^G «smaler causes the overall picture to be bluish. The white balance gain is now corrected using the above method of the embodiment of the invention:

(a) Recognizing the face area of the blue-blue image processed by the automatic white balance, the calculated RGB statistical values of the face area are: =137.4, G ₀ =114.4, B ₀ =115.6;

Calculate the brightness value of the face area. :

Y ₀ =0.114*137.4+(1-0.114-0.299)*114.4+0.299*115.6=185.792 Calculate the brightness value of the ideal face model:

=0.114*172.5+(1-0.114-0.299)*117.3+0.299*99.3=118.2108

(b) The brightness value of the face area. Calculate the new white balance gain for each channel by the brightness value greater than the ideal face model: G _R = (R _l /R ₀ )^ ^,r " ⁾ *G _R =(172.5/137.4)( ¹¹⁸² ^S5792) *3.48 =4.02

G _G = (G ₁ /G ₀ ^l/7 ° ⁾ *G _G = (117.3/114.4) ^{(n82108/185792)} *l=1.016

(C) According to the new white balance gain ', G _G , (3⁄4 ' of the three channels of red, green and blue RGB, white balance processing is performed on the blue image, and it can be seen that the red channel has a new white balance gain ratio. Originally improved, the new white balance gain of the blue channel is lower than the original. Therefore, the color after the white balance processing is closer to the human eye. Example 2, assume the image after the automatic white balance processing The overall yellowing, the current RGB gain are: ^G «=3.963, ^G = l, ^G »=1.518, the RGB values of the ideal face model are: = 172.5, ^G i = 117.3, ^B , = 99.3; It is known that if ^G « is corrected to 3.8, ^{G is} corrected to 1, and ^G « is corrected to 1.8, the color of the image is close to the normal color seen by the human eye. Therefore, it can be seen that the image of the automatic white balance processing has a small blue channel gain. The overall picture is yellowish, and now the white balance gain is corrected using the above method: (a) Recognizing the face region of the yellowed image subjected to the automatic white balance processing, the calculated RGB statistical values of the face region are: = 174.1761, G ₀ = 118.649, B ₀ = 82.4667;

Calculate the brightness value of the face area. :

7 ₀ = 0.114*174.1761+(1-0.114-0.299)* 118.649+0.299*82.4667=114.160

Calculate the brightness value of the ideal face model:

=0.114*172.5+(1-0.114-0.299)*117.3+0.299*99.3=118.2108

(b) The brightness value of the face area. Calculating a new white balance gain for each channel, less than the brightness value of the ideal face model, where Y _max = 255;

G _R = (172.5/174.1761) ⁽⁽²⁵⁵ - ¹¹⁸ - ^{2108) / (255} - ^114160)) * 3.963

=3.926;

G _a = ( /Go W)*G _G= (117.3/118.649) ⁽⁽²⁵⁵ - ^{11S )/(255} - ^114160)) *1

=0.989;

_{Gs = ( ι} / = (99.3/82.4667) * 1.518

=1.818;

(c) According to the new white balance gains ', G _G , (3⁄4 ' of the three channels of red, green and blue RGB, white balance processing is performed on the blue image, and it can be seen that the blue channel has a new white balance gain. As shown in FIG. 2, the embodiment of the present invention provides a system for calibrating the color of an image, and the system includes:

The face recognition module 201 is configured to perform face recognition on the image subjected to the automatic white balance processing, and if the recognition is successful, determine a face region, and calculate a red, green and blue RGB statistical value of the face region;

The gain calculation module 202 is configured to calculate a corrected white balance gain of three channels of red, green and blue RGB required to correct the RGB statistical value of the face region to the RGB value of the ideal face model; the white balance processing module 203, It is arranged to perform white balance processing on the image subjected to the automatic white balance processing according to the corrected white balance gain of the three channels of red, green and blue RGB.

The system also includes the following features: The face area is a rectangular area or a circular area including a face image.

The face recognition module calculates the red, green, and blue RGB statistics of the face region by: summing the red, green, and blue RGB values of all the pixels of the face region and averaging the image to obtain the face. The red, green and blue RGB average of the area.

The gain calculation module calculates the corrected white balance gain of the three channels of red, green, and blue RGB required to correct the RGB statistical value of the face region to the RGB value of the ideal face model as follows:

(a) calculating a brightness value of the face region according to an RGB average value of the face region;

(b) calculating the brightness value of the ideal face model according to the RGB value of the ideal face model;

(c) comparing the brightness value of the face area with the brightness value of the ideal face model, and calculating the gain correction factor of each of the three channels of red, green and blue RGB according to the brightness comparison result, the red, green and blue RGB The current white balance gain of each channel is multiplied by the channel's gain correction factor to obtain the corrected white balance gain for that channel.

3⁄4 = (W ^) ;

a _a = . strict);

When the brightness value of the face area. When the luminance value J of the ideal face model is smaller than the luminance of the ideal face model, the difference between the luminance maximum value and the luminance value of the ideal face model is divided by the quotient obtained by dividing the primary color component of the ideal face model by the corresponding primary color component value of the face region. a quotient obtained by the difference between the luminance maximum value and the luminance value of the face region as an index, and the base and the exponential constructed power are used as the channel corresponding to the primary color Gain correction factor; the gain correction factor of the red channel " _R , the gain correction factor of the green channel" _σ , the gain correction factor of the blue channel are as follows:

a _R = strict leg u;

Where is the red component value of the face region, R, which is the red component value of the ideal face model, G. The green component value of the face region is the green component value of the ideal face model, S. It is the blue component value of the face region, A is the blue component value of the ideal face model, Y _max is the brightness maximum value, and the Y dish is preset. The method and system for calibrating the color of an image provided by the above embodiment, by performing face recognition on the image subjected to the automatic white balance processing, calculating the RGB value required to correct the RGB statistical value of the face region to the RGB value of the ideal face model The corrected white balance gain of the three channels of red, green and blue RGB, the image is re-white balanced according to the corrected white balance gain, and the above method and system can make the color of the processed image appear closer to the human eye. The color that improves the image quality.

One of ordinary skill in the art will appreciate that all or a portion of the above steps may be accomplished by a program instructing the associated hardware, such as a read-only memory, a magnetic disk, or an optical disk. Optionally, all or part of the steps of the foregoing embodiments may also be implemented by using one or more integrated circuits. Accordingly, each module/unit in the foregoing embodiment may be implemented in the form of hardware, or may use software functions. The form of the module is implemented. The invention is not limited to any specific form of combination of hardware and software.

It is to be understood that the invention may be embodied in various other embodiments without departing from the spirit and scope of the invention.

Industrial applicability

The embodiment of the invention can make the color of the processed image appear closer to the color seen by the human eye and improve the image quality.

Claims

claims

1. A method of calibrating image color, including:

Perform face recognition on the image that has been processed by automatic white balance. If the recognition is successful, determine the face area and calculate the red, green, and blue RGB statistical values of the face area;

Calculate the corrected white balance gain of the three red, green, and blue RGB channels required to correct the RGB statistical value of the face area to the RGB value of the ideal face model; and

The image that has undergone automatic white balance processing is re-white balanced according to the corrected white balance gain of the three RGB channels.

2. The method of claim 1, wherein:

Calculate the red, green and blue RGB statistical values of the face area, including:

The red, green, and blue RGB values of all pixels in the human face area are summed and averaged to obtain the average red, green, and blue RGB values of the human face area.

3. The method of claim 2, wherein:

Calculate the corrected white balance gain of the three red, green, and blue RGB channels required to correct the RGB statistical value of the face area to the RGB value of the ideal face model, including:

Calculate the brightness value of the face area according to the RGB average value of the face area;

Calculate the brightness value of the ideal face model based on the RGB values of the ideal face model; and

Compare the brightness value of the face area with the brightness value of the ideal face model, calculate the gain correction factor for each of the three red, green, and blue RGB channels based on the brightness comparison results, and divide each of the red, green, and blue RGB channels into The current white balance gain is multiplied by the gain correction factor of the channel to obtain the corrected white balance gain of the channel.

4. The method of claim 3, wherein:

The gain correction factor for each of the three red, green, and blue RGB channels is calculated based on the brightness comparison results, including:

When the brightness value of the face area. When it is greater than or equal to the brightness value of the ideal face model, the ideal face model will The quotient obtained by dividing the primary color component of the face model by the corresponding primary color component value of the face area is used as the base. The quotient obtained by dividing the brightness value of the ideal face model by the brightness value of the face area is used as the index. The base and the resulting The power of the exponential structure is used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel < _R , the gain correction factor of the green channel < _σ , and the gain correction factor < ¾ of the blue channel are as follows:

¾ = (W ^) ;

a _a = . strict);

When the brightness value of the face area. When it is less than the brightness value of the ideal face model, the quotient obtained by dividing the base color component of the ideal face model by the corresponding base color component value of the face area is used as the base, and the difference between the maximum brightness value and the brightness value of the ideal face model is divided by The quotient obtained by the difference between the maximum brightness value and the brightness value of the face area is used as an index, and the power constructed by the base number and the index is used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel « _R , The gain correction factor o _Q of the green channel and the gain correction factor of the blue channel are as follows:

a _R = ( /T^uu;

Among them, is the red component value of the face area, R, is the red component value of the ideal face model, G. is the green component value of the face area, is the green component value of the ideal face model, S. is the blue component value of the face area, A is the blue component value of the ideal face model, Y _max is the maximum brightness value, and the Y max is preset.

5. The method of claim 1, wherein:

The face area is a rectangular area or a circular area containing a face image.

6. A system for calibrating image color, including:

A face recognition module, which is configured to perform face recognition on the image that has undergone automatic white balance processing. If the recognition is successful, determine the face area and calculate the red, green, and blue RGB statistical values of the face area; a gain calculation module, which Set to calculate and correct the RGB statistical value of the face area to the ideal face The corrected white balance gain of the three red, green, and blue RGB channels required by the RGB value of the model; and a white balance processing module, which is configured to perform the processing according to the corrected white balance gain of the three red, green, and blue RGB channels. Images processed with automatic white balance are re-white balanced.

7. The system of claim 6, wherein:

The face recognition module calculates the red, green, and blue RGB statistical values of the face area in the following way: sum the red, green, and blue RGB values of all pixels in the face area and take the average to obtain the red, green, and blue RGB values of the face area. Red, green and blue RGB average.

8. The system of claim 7, wherein:

The gain calculation module calculates the corrected white balance gain of the three red, green, and blue RGB channels required to correct the RGB statistical value of the face area to the RGB value of the ideal face model in the following way: According to the RGB average value of the face area Calculate the brightness value of the face area;

9. The system of claim 7, wherein:

The gain calculation module calculates the gain correction factor for each of the three red, green, and blue RGB channels based on the brightness comparison results as follows:

When the brightness value of the face area. When it is greater than or equal to the brightness value of the ideal face model, the quotient obtained by dividing the base color component of the ideal face model by the corresponding base color component value of the face area is used as the base, and the brightness value of the ideal face model is divided by the base color component of the face area. The quotient obtained from the brightness value is used as an index, and the base and the power constructed by the index are used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel « _R , the gain correction factor of the green channel « _σ , blue The gain correction factors for the color channels are as follows: When the brightness value of the face area. When it is less than the brightness value of the ideal face model, the quotient obtained by dividing the base color component of the ideal face model by the corresponding base color component value of the face area is used as the base, and the difference between the maximum brightness value and the brightness value of the ideal face model is divided by The quotient obtained by the difference between the maximum brightness value and the brightness value of the face area is used as an index, and the power constructed by the base number and the index is used as the gain correction factor of the channel corresponding to the primary color; the gain correction factor of the red channel « _R , The gain correction factor o _Q of the green channel and the gain correction factor of the blue channel are as follows:

a _R = Yanji u;

a _G = ( /GH - w) ;

Among them, is the red component value of the face area, R, is the red component value of the ideal face model, G. is the green component value of the face area, is the green component value of the ideal face model, S. is the blue component value of the face area, A is the blue component value of the ideal face model, Y _max is the maximum brightness value, as described

10. The system of claim 6, wherein:

The face area is a rectangular area or a circular area containing a face image.