CN106878695A - White balance processing method, device and computer equipment - Google Patents

Abstract

The invention relates to a white balance processing method, a white balance processing device and computer equipment. The method comprises the following steps: carrying out face recognition on the image, and recognizing a face region and a background region except the face region; acquiring a reference area from the face area, wherein the reference area is a white area in a real object; carrying out white balance processing on the whole image according to the reference area; or, performing white balance processing on the face region according to the reference region; and carrying out white balance processing on the regions except the human face region according to the background region. According to the embodiment of the invention, the white balance processing can be carried out on the image according to the specific reference area in the face area of the image, so that the accuracy of the white balance is improved, and the user experience is improved.

Description

White balance processing method, device and computer equipment

technical field

The invention relates to image processing technology, in particular to a white balance processing method, device and computer equipment.

Background technique

Color Temperature is a measure of the light color of a light source, and the unit is K (Kelvin). The human eye perceives the brightest objects as white at any color temperature. The photos taken by the camera at different color temperatures show different colors, for example, the photos under the D65 light source are bluish, while the photos under the A light are yellowish. Indoor light sources are often more complicated, and the color temperature of incandescent lamps and fluorescent lamps is not very standard. Therefore, shooting portraits indoors often leads to abnormal skin tones, yellowish or bluish.

With the development of image processing technology, people have higher and higher requirements for images, and usually post-optimize images to obtain better visual effects. Automatic White Balance (AWB) is widely used in the processing of portrait pictures containing human faces. The essence of White Balance (WB) is to make white objects appear white under any color light source. White balance makes the color of the captured image become the normal color seen by human eyes through color correction. The photo read from the photosensitive chip is called the original picture, and the automatic white balance color correction is performed on the original picture to achieve the white balance effect.

However, in some scenarios, the effect of automatic white balance is still different from the normal color seen by the human eye, and the problem of color shift will occur, and the user experience is not good.

Contents of the invention

Embodiments of the present invention provide a white balance processing method, device, and computer equipment, which can more accurately adjust the white balance of an image and improve user experience.

A method for white balance processing, comprising:

Perform face recognition on the image, identify the face area and the background area except the face area;

Obtaining a reference area from the face area, wherein the reference area is a white area in the real object;

Perform white balance processing on the entire image according to the reference area; or perform white balance processing on the face area according to the reference area; perform white balance processing on the area other than the face area according to the background area .

A device for white balance processing, comprising:

The face recognition module is used to identify the face area and the background area except the face area;

An acquisition module, configured to acquire a reference area in the face area, wherein the reference area is a white area in the real object;

A white balance processing module, configured to perform white balance processing on the entire image according to the reference area; or, perform white balance processing on the human face area according to the reference area and perform white balance processing on the human face area according to the background area White balance processing is performed on the other areas.

A computer device, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, and the processor implements the following steps when executing the program:

Perform face recognition on the image, and obtain the face area and the background area except the face area;

Acquiring a reference area in the face area, wherein the light reflected from the reference area to the human eye in the real object has a certain brightness and the proportions of blue, green, and red colors in the light are the same;

Perform white balance processing on the entire image according to the reference area; or, perform white balance processing on the face area according to the reference area; determine a background area based on the face area, and the background area is the person except the person Areas other than the face area; performing white balance processing on the area other than the face area according to the background area.

Compared with the traditional automatic white balance method, the above white balance processing method can perform white balance processing on the image according to a specific reference area in the face area of the image, thereby improving the accuracy of the white balance and improving user experience.

Description of drawings

FIG. 1 is a schematic diagram of the internal structure of a terminal in an embodiment;

Fig. 2 is a flowchart of a method for white balance processing in an embodiment;

Fig. 3 is a flowchart of a method for white balance processing in another embodiment;

Fig. 4 is a flowchart of obtaining a reference area in the face area in an embodiment;

Fig. 5 is a flow chart of performing white balance processing on an image according to the reference area in an embodiment;

Fig. 6 is a flow chart of performing white balance processing on an image according to the background area in an embodiment;

FIG. 7 is a structural frame diagram of a device for white balance processing in an embodiment;

Fig. 8 is a structural frame diagram of an acquisition module in a device for white balance processing in an embodiment;

FIG. 9 is a structural frame diagram of a white balance processing module in a white balance processing device in an embodiment;

Fig. 10 is a flowchart of the steps implemented when the processor of the computer device executes the computer program in one embodiment.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

It can be understood that the terms "first", "second" and the like used in the present invention can be used to describe various elements herein, but these elements are not limited by these terms. These terms are only used to distinguish one element from another element. For example, a first statistical unit could be termed a second statistical unit, and, similarly, a second statistical unit could be termed a first statistical unit, without departing from the scope of the present invention. Both the first statistical unit and the second statistical unit are statistical units, but they are not the same statistical unit.

Fig. 1 is a schematic diagram of the internal structure of a terminal in an embodiment. As shown in FIG. 1 , the terminal includes a processor 101 connected through a system bus, a non-volatile storage medium 102 , an internal memory 103 , a network interface 104 , a display screen 105 , a camera 106 , and an image sensor 107 . Wherein, the non-volatile storage medium of the terminal stores an operating system, and further includes a white balance processing device 108, and the white balance processing device 108 is used to implement a white balance processing method. The processor 101 is used to provide computing and control capabilities to support the operation of the entire terminal. The internal memory 103 of the terminal may store computer-readable instructions, and when the computer-readable instructions are executed by the processor 101, the processor 101 may execute a white balance processing method. The network interface 104 is used for network communication with the server, such as sending a news data access request to the server, receiving news data returned by the server, and so on. The display screen 105 of the terminal may be a liquid crystal display screen or an electronic ink display screen or the like. The terminal may be a mobile phone, a tablet computer, a digital camera, and the like. Those skilled in the art can understand that the structure shown in Figure 1 is only a block diagram of a partial structure related to the solution of this application, and does not constitute a limitation on the terminals to which the solution of this application is applied. Specific terminals may include More or fewer components are shown in the figures, or certain components are combined, or have different component arrangements.

As shown in FIG. 2 , in one embodiment, a white balance processing method is provided, and this embodiment is described by taking this method applied to the terminal in FIG. 1 above as an example. The method specifically includes the following steps:

Step 202, face recognition is performed on the image, and a face area and a background area other than the face area are obtained.

The image to be processed is divided into regions, and various face recognition algorithms can be used to identify the face region, and the region in the image other than the face region is defined as the background region.

The face recognition method can be based on the face recognition method of principal component analysis (PCA), from a statistical point of view, to find the basic elements of the face image distribution, that is, the eigenvector of the covariance matrix of the face image sample set, In this way, the face image is approximately represented. These feature vectors are called Eigenfaces. The face recognition method can also perform skin color detection through color analysis to locate the face, use the color characteristics of the facial skin to establish a new color coordinate system, and realize the positioning of the face by separating the skin color from the image. The face recognition method can also be a deformed template method, which uses an ellipse to approximate the head contour and refines it through iteration. The face recognition method may also use the Adaboost algorithm. Preferably, a dynamic threshold can be used in the Adaboost algorithm to further accelerate the speed of face recognition. The face recognition algorithm may also use other algorithms capable of quickly recognizing face regions, and the face recognition algorithm is not specifically limited in the embodiment of the present invention.

Further, face recognition is performed on the image, and obtaining the face area and the background area other than the face area may include: performing face recognition on an image with a face, and determining a rectangular frame corresponding to the face; Extracting the contour of the human face; using the contour of the human face as the face area, and other areas in the image as the background area. Wherein, extracting the contour of the human face in the rectangular frame may use an active shape model or an active appearance model to extract the contour of the human face.

Step 204, acquiring a reference area in the face area, wherein the reference area is a white area in the real object.

The reference area in the face area is located by the face recognition algorithm. The reference area can be understood as the white area seen by the human eyes in the real object, or can be understood as the white area in the real human face seen by the eyes. Because white refers to the visual reaction formed by the light reflected into the human eye due to the same proportion of blue, green and red colors and a certain brightness, the white in the real object does not contain the brightness of the color component. The reference area can be the white area of the eye, or the tooth area.

Step 206, perform white balance processing on the entire image according to the reference area.

The whole image is white balanced according to the red, green and blue RGB vectors of the reference area (eye white area or tooth area).

In one embodiment, referring to FIG. 3 , step 208 may also be used instead of step 206 . Step 208, perform white balance processing on the face area according to the reference area; perform white balance processing on the areas other than the face area according to the background area. That is, perform white balance processing on the face area according to the color vector of the reference area; meanwhile, perform white balance processing on the areas other than the face area according to the color vector of the background area.

Compared with the traditional automatic white balance processing of the entire image based on the color vector of the entire image, the white balance processing method in the embodiment of the present invention recognizes the face area and the background other than the face area through face recognition At the same time, it can also locate the reference area in the face area, perform automatic white balance processing on the entire image according to a specific reference area, or perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and perform automatic white balance processing on the face area according to a specific reference area, and according to the background area Perform white balance processing on the area outside the face area in the entire image, thereby improving the accuracy of white balance and improving user experience.

In one embodiment, referring to FIG. 4, step 204, obtaining a reference area from the face area includes:

Step 402, locate the eye area in the face area.

After the face area is identified according to the face recognition method, the eyes in the face area are identified or positioned. The recognition of human eyes includes edge feature analysis method, Hough transform method and deformation template method. Human eye recognition can also use the median filter and histogram equalization method to remove the influence of noise and light on the image, and then integrally project the image to reduce it to the eye area of the face, and then do a leveling in the obtained eyebrow area Integral projections to find the vertical position of the eyes. Finally, the part with the highest degree of matching along the vertical direction using the human eye template is the human eye area to be located. In the embodiment of the present invention, no specific limitation is imposed on the human eye recognition method.

Step 404, extracting the white area of the eye from the eye area and defining the white area of the eye as a reference area.

Eye regions are located or identified according to step 402 . The eye area includes eye white, pupil, and part of the eyelid. The eye white area is extracted from the eye area, and the eye white area is defined as a reference area.

Under normal circumstances, the whites of the human eye are white. In terms of color composition, it can be considered that the ratio of the red, green and blue RGB color components of the white part of the eyes is 1:1:1. The white area of the eyes is defined as the reference area, and the red, green, and blue RGB color components of the reference area are restored to the color of the whites of the human eyes under normal circumstances, thereby realizing automatic whitening processing of the face area or the entire image.

When extracting the white area of the eye, the brightness histogram of the eye area is calculated separately, and the white area of the eye is located according to the brightness histogram.

In one embodiment, the step of extracting the eye white area from the eye area and defining the eye white area as a reference area specifically includes:

Obtain the brightness value of each pixel in the eye region; obtain the average brightness value of the eye region according to the brightness value of each pixel point in the eye region; extract the white region of the eye, and select the region in which the pixel brightness in the eye region is greater than the average brightness value It is defined as the white area of the eye, that is, the brightness of the pixels in the white area of the eye is greater than the average brightness.

In one embodiment, step 204, acquiring a reference area in the face area may also be:

In the human face area, a tooth area is located and extracted, and the tooth area is defined as a reference area. After the face area is identified according to the face recognition method, the teeth in the face area are identified or positioned.

Under normal circumstances, teeth are also white. In terms of color composition, it can be considered that the ratio of the red, green and blue RGB color components of the teeth is 1:1:1. The tooth area is defined as a reference area, and the red, green, and blue RGB color components of the reference area are restored to the white color of the human eye under normal circumstances, thereby realizing automatic whitening processing of the face area or the entire image.

In one embodiment, the eye white area and the tooth area are recognized simultaneously in the image through the method of face recognition, and the eye white area is firstly defined as a reference area, and white balance processing is performed on the entire image or the face area with the eye white area as a reference. In one embodiment, the weights of the red, green and blue RGB color vectors of the eye white area and the tooth area can also be considered comprehensively, and the eye white area and the tooth area can be defined as reference areas, and then white balance processing can be performed on the entire image or the face area.

In one embodiment, referring to FIG. 5, step 206, performing white balance processing on the entire image according to the reference area, includes:

Step 502, counting the average value or maximum value of the red, green and blue RGB color components of all pixels in the reference area.

Mention the color components of all pixels in the reference area, and the color of each pixel is represented by three primary color components: Red, Green, and Blue. Take the average value of the red, green and blue RGB color components of all pixels in the reference area to obtain the average value of the red, green and blue RGB color components of the reference area Alternatively, select the maximum value of the red, green, blue RGB color components of all pixels in the reference area to obtain the maximum value of the red, green, blue RGB color components (R _max , G _max , B _max ) in the reference area.

Step 504, calculating correction adjustment factors for the red R channel and the blue B channel according to the mean or maximum value of the red, green and blue RGB color components.

The input data for white balance processing is a two-dimensional vector of (red, blue), green is the reference color, and there is no need to participate in white balance.

In one example, according to the mean value of the red, green and blue RGB color components corresponding to the reference area Calculate the correction adjustment factors for the red R channel and blue B channel. Wherein, the red R channel correction adjustment factor K _R is the quotient of the green channel component mean value divided by all the red R channel components R in the reference area. The blue B channel correction adjustment factor K _B is the quotient of the green channel component mean value divided by all blue B channel components B in the reference area.

In one example, the correction adjustment factors of the red R channel and the blue B channel are calculated according to the maximum values of the red, green and blue RGB color components (R _max , G _max , B _max ) corresponding to the reference area. Wherein, the red R channel correction adjustment factor l _R is the quotient of the maximum value of the green channel component divided by the maximum value of the red channel in the reference area. The blue B channel correction adjustment factor l _B is the quotient of the maximum value of the green channel component divided by the maximum value of the blue channel in the reference area.

Step 506, performing white balance processing on the entire image according to the correction adjustment factors of the red R channel and the blue B channel.

Bring the correction adjustment factors of the red R channel and blue B channel calculated in step 504 into the corresponding white balance processing model to realize white balance processing on the entire image or face area, and quickly perform white balance on the face area , to improve white balance efficiency.

In other embodiments, when performing white balance processing on the entire image according to the reference area, the white balance processing method can also be a simple grayscale world algorithm (GW) and a total perfect reflection algorithm (PR) quadrature combination algorithm (QCGP ), Standard Deviation Weighted Gray World Algorithm (SDWGW), Luminance Weighted Gray World Algorithm (LWGW), Standard Deviation Luminance Gray World Algorithm (SDLWGW), and Luminance Weighted Gray World Algorithm and Total Perfect Reflection Algorithm (PR) Orthogonal combination algorithm (QCLWG P) and so on.

In one embodiment, step 208, performing white balance processing on the face area according to the reference area; performing white balance processing on the area other than the face area according to the background area, including:

According to steps 502 to 506, after performing white balance processing on the face area according to the reference area, then performing white balance processing on the areas other than the face area according to the background area.

Wherein, according to the background area, the white balance processing is performed on the area other than the face area, referring to FIG. 6, including:

Step 602, counting the average value or maximum value of the red, green and blue RGB color components of all pixels in the background area.

The color components of all pixels in the background area are mentioned, and the color of each pixel is represented by three primary color components of red, green, and blue. Take the average value of the red, green and blue RGB color components of all pixels in the background area to obtain the average value of the red, green and blue RGB color components of the reference area Alternatively, select the maximum value of the red, green, blue RGB color components of all pixels in the background area to obtain the maximum value of the red, green, blue RGB color components (R _max , G _max , B _max ) in the reference area.

Step 604, calculating correction adjustment factors for the red R channel and the blue B channel according to the mean or maximum value of the red, green and blue RGB color components.

The input data for white balance processing is a two-dimensional vector of (red, blue), green is the reference color, and there is no need to participate in white balance.

In one example, according to the mean value of the red, green and blue RGB color components corresponding to the reference area Calculate the correction adjustment factors for the red R channel and blue B channel. Wherein, the red R channel correction adjustment factor K _R is the quotient of the green channel component mean value divided by all the red R channel components R in the reference area. The blue B channel correction adjustment factor K _B is the quotient of the green channel component mean value divided by all blue B channel components B in the reference area.

In one example, the correction adjustment factors of the red R channel and the blue B channel are calculated according to the maximum values of the red, green and blue RGB color components (R _max , G _max , B _max ) corresponding to the reference area. Wherein, the red R channel correction adjustment factor l _R is the quotient of the maximum value of the green channel component divided by the maximum value of the red channel in the reference area. The blue B channel correction adjustment factor l _B is the quotient of the maximum value of the green channel component divided by the maximum value of the blue channel in the reference area.

Step 606, performing white balance processing on the background area image according to the correction adjustment factors of the red R channel and the blue B channel.

The correction adjustment factors of the red R channel and blue B channel corresponding to the background area calculated in step 604 are brought into the corresponding white balance processing model to realize white balance processing on the background area, and realize rapid adjustment of the image except for the face area. The hair and environment can be white balanced to improve the accuracy of the white balance, reduce the difference between the image and the real scene, and be able to perform white balance processing more accurately and improve the user experience.

Fig. 7 is a structural block diagram of a device for white balance processing in an embodiment. As shown in Figure 8, a device for white balance processing includes:

A face recognition module 710, configured to identify a face area and a background area other than the face area;

An acquisition module 720, configured to acquire a reference area in the face area, wherein the reference area is a white area in the real object;

The white balance processing module 730 is configured to perform white balance processing on the entire image according to the reference area; or, perform white balance processing on the face area according to the reference area and perform the face removal process on the face area according to the background area. Areas outside the area are white-balanced.

The above-mentioned white balance processing device can recognize the face in the image when performing white balance processing on the image, and recognize the face area and the background area other than the face area. Obtain a reference area in the face area, and perform white balance processing on the entire image according to the reference area. Or, perform white balance processing on the face area according to the reference area; perform white balance processing on the areas other than the face area according to the background area. The embodiment of the present invention performs white balance processing on the image according to a specific reference area in the face area of the image, thereby improving the accuracy of the white balance and improving user experience.

In one embodiment, the face recognition module 710 can divide the image to be processed into regions. Various face recognition algorithms can be used to identify the face region, and the region in the image other than the face region is defined as the background region. The face recognition module 710 can be based on the face recognition method of principal component analysis (PCA), from a statistical point of view, to find the basic elements of the face image distribution, that is, the eigenvectors of the covariance matrix of the face image sample set , which approximates the face image. The face recognition module 710 can also perform skin color detection through color analysis to locate the face, use the color characteristics of the facial skin to establish a new color coordinate system, and realize the positioning of the face by separating the skin color from the image. The face recognition module 710 can also be a deformed template method, which uses an ellipse to approximate the head contour, and refines it through iteration. The face recognition module 710 may also use other algorithms capable of quickly recognizing face regions, and the face recognition algorithm is not specifically limited in this embodiment of the present invention.

In one embodiment, the obtaining module 720 can obtain the reference area from the face recognition module 710 . The reference area can be the white area of the eye, or the tooth area. The reference area can be understood as the white area seen by the human eyes in the real object, or can be understood as the white area in the real human face seen by the eyes. Because white refers to the visual reaction formed by the light reflected into the human eye due to the same proportion of blue, green and red colors and a certain brightness, the white in the real object does not contain the brightness of the color component.

In one embodiment, referring to FIG. 8 , the acquiring module 720 includes a positioning unit 721 and an extracting unit 723 . Wherein, the positioning unit 721 is configured to locate the eye area or the tooth area in the human face area. The extracting unit 723 is used for extracting the eye white area in the eye area or for extracting the tooth area.

Through the positioning unit 721 and the extraction unit 723, the eye area or the tooth area can be identified from the human face area, and then the reference area can be extracted from the eye area or the tooth area through the extraction unit 723, wherein the reference area can be the white area of the eye or the tooth area.

In one embodiment, the acquisition module 720 further includes a calculation unit 725 and an selection unit 727 . Wherein, the calculation unit 725 is configured to calculate the luminance value of each pixel in the eye area to obtain an average luminance value of the eye area. The selection unit 727 is configured to select pixels whose brightness in the eye area is greater than the average brightness value, and the set of selected pixels is the white area of the eye.

During the process of extracting the white of the eye, the calculation unit 725 calculates the brightness histogram of the eye area. According to the brightness histogram, obtain the average brightness value of the eye area. The selection unit 727 selects pixels whose brightness in the eye region is greater than the average brightness value, and the set of selected pixels is the eye white region.

In one embodiment, referring to FIG. 9 , the white balance processing module 730 includes a first statistical unit 731 , a first gain correction factor acquisition unit 732 and a first correction unit 733 . Among them, the first statistics unit 731 is used to calculate the average value or maximum value of the red, green and blue RGB color components of all pixels in the reference area. The first gain correction factor acquisition unit 732 is configured to calculate the correction adjustment factors of the red R channel and the blue B channel according to the mean or maximum value of the red, green and blue RGB color components in the reference area. The first correction unit 733 is configured to perform white balance processing on the entire image according to the correction adjustment factors of the red R channel and the blue B channel of the reference area. The first correction unit 733 implements white balance processing on the entire image or face area according to the correction adjustment factors of the red R channel and blue B channel corresponding to the reference area obtained by the first gain correction factor acquisition unit 732, and realizes fast correction Perform white balance on the face area to improve white balance efficiency.

In one embodiment, the white balance processing module 730 further includes a second statistical unit 736 , a second gain correction factor acquisition unit 737 and a second correction unit 738 . Wherein, the second statistics unit 736 is used to calculate the mean value or maximum value of the red, green and blue RGB color components of all pixels in the background area. The second gain correction factor acquisition unit 737 is configured to calculate the correction adjustment factors of the red R channel and the blue B channel according to the mean value or maximum value of the red, green and blue RGB color components corresponding to the background area. The second correction unit 738 is configured to perform white balance processing on the background region image according to the correction adjustment factors of the red R channel and the blue B channel corresponding to the background region. The second correction unit 738 performs white balance processing on the background area according to the correction adjustment factors of the red R channel and the blue B channel of the background area acquired by the second gain correction factor acquisition unit 737, so as to quickly correct the hair in the image except for the face area. , The environment performs white balance, improves the accuracy of white balance, reduces the difference between the image and the real scene, can perform white balance processing more accurately, and improves user experience.

The division of each module in the above-mentioned white balance processing device is only for illustration. In other embodiments, the white balance processing device can be divided into different modules according to needs, so as to complete all or part of the above-mentioned white balance processing device Function.

Fig. 10 is a flowchart of the steps implemented when the processor of the computer device executes the computer program in one embodiment. As shown in Figure 10, a kind of computer equipment comprises memory, processor and the computer program (instruction) that is stored on memory and can run on processor, and processor realizes following steps when executing program:

Step 1002, face recognition is performed on the image, and a face area and a background area other than the face area are obtained.

The image to be processed is divided into regions, and various face recognition algorithms can be used to identify the face region, and the region in the image other than the face region is defined as the background region.

The face recognition method can be based on the face recognition method of principal component analysis (PCA), from a statistical point of view, to find the basic elements of the face image distribution, that is, the eigenvector of the covariance matrix of the face image sample set, In this way, the face image is approximately represented. These feature vectors are called Eigenfaces. The face recognition method can also perform skin color detection through color analysis to locate the face, use the color characteristics of the facial skin to establish a new color coordinate system, and realize the positioning of the face by separating the skin color from the image. The face recognition method can also be a deformed template method, which uses an ellipse to approximate the head contour and refines it through iteration. The face recognition method may also use the Adaboost algorithm. Preferably, a dynamic threshold can be used in the Adaboost algorithm to further accelerate the speed of face recognition. The face recognition algorithm may also use other algorithms capable of quickly recognizing face regions, and the face recognition algorithm is not specifically limited in the embodiment of the present invention.

Further, face recognition is performed on the image, and obtaining the face area and the background area other than the face area may include: performing face recognition on an image with a face, and determining a rectangular frame corresponding to the face; Extracting the contour of the human face; using the contour of the human face as the face area, and other areas in the image as the background area. Wherein, extracting the contour of the human face in the rectangular frame may use an active shape model or an active appearance model to extract the contour of the human face.

Step 1004, obtain a reference area in the face area, wherein the light reflected from the reference area in the real object to the human eye has a certain brightness and the proportions of blue, green, and red colors in the light are the same .

The reference area in the face area is located by the face recognition algorithm. The reference area can be the white area of the eye, or the tooth area. The reference area can be understood as the white area seen by the human eyes in the real object, or can be understood as the white area in the real human face seen by the eyes. Because white refers to the visual reaction formed by the light reflected into the human eye due to the same proportion of blue, green and red colors and a certain brightness, the white in the real object does not contain the brightness of the color component.

Step 1006, perform white balance processing on the entire image according to the reference area. Or perform white balance processing on the face area according to the reference area; perform white balance processing on the areas other than the face area according to the background area.

The whole image is white balanced according to the red, green and blue RGB vectors of the reference area (eye white area or tooth area). Or, perform white balance processing on the face area according to the color vector of the reference area; meanwhile, perform white balance processing on the areas other than the face area according to the color vector of the background area.

When the processor in the above-mentioned computer device executes the program, it recognizes the face area and the background area other than the face area through face recognition, and at the same time, it can also locate the reference area in the face area, according to the specific reference area. Perform automatic white balance processing on the entire image, or perform automatic white balance processing on the face area according to a specific reference area, and perform white balance processing on the area outside the face area in the entire image according to the background area, thereby improving the accuracy of the white balance. Accuracy, improve user experience.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be realized through computer programs to instruct related hardware, and the programs can be stored in a non-volatile computer-readable storage medium When the program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, the storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM) and the like.

The above-mentioned embodiments only express several implementation modes of the present invention, and the description thereof is relatively specific and detailed, but should not be construed as limiting the patent scope of the present invention. It should be pointed out that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent for the present invention should be based on the appended claims.

Claims (13)

1. a kind of method that white balance is processed, it is characterised in that including：

Recognition of face is carried out to image, human face region and the background area in addition to the human face region is recognized；

Reference zone is obtained from the human face region, wherein, the reference zone is the white portion in material object；

White balance treatment is carried out to whole image according to the reference zone；Or, according to the reference zone to the face Region carries out white balance treatment；White balance treatment is carried out to the region in addition to human face region according to the background area.

2. the method that white balance according to claim 1 is processed, it is characterised in that described to be obtained from the human face region Reference zone, including：

Eye areas are positioned in the human face region；

White of the eye region is extracted from the eye areas and the white of the eye region is defined for reference zone.

3. the method that white balance according to claim 2 is processed, it is characterised in that extract the white of the eye from the eye areas Region simultaneously defines the white of the eye region for reference zone, including：

Obtain the brightness value of each pixel of the eye areas；

The brightness value of each pixel according to the eye areas obtains the luminance mean value of the eye areas；

White of the eye region is extracted, the brightness of the pixel in the white of the eye region is more than the luminance mean value.

4. the method that white balance according to claim 1 is processed, it is characterised in that according to the reference zone to whole figure As carrying out white balance treatment, including：

Count the RGB RGB color component average or maximum of the reference zone all pixels point；

Correction according to the reference zone RGB RGB color component average or the red R passages of maximum value calculation and blue channel B is adjusted The section factor；

Correction regulatory factor according to the red R passages of the reference zone and blue channel B carries out white balance treatment to whole image.

5. the method that white balance according to claim 4 is processed, it is characterised in that the red R passages of calculating and blue channel B Correction regulatory factor, including：

The green channel components average is respectively divided by all red R channel components of the reference zone, blue channel B component to deserved To red R passages, the correction regulatory factor of blue channel B；Or

The green channel components maximum is respectively divided by the red passage maximum of the reference zone, blue channel maximum to deserved To red R passages, the correction regulatory factor of blue channel B.

6. the method for white balance according to claim 1 treatment, it is characterised in that it is described according to the background area to removing Region beyond human face region carries out white balance treatment, including：

Count the RGB RGB color component average or maximum of the background area all pixels point；

Correction according to the background area RGB RGB color component average or the red R passages of maximum value calculation and blue channel B is adjusted The section factor；

The correction regulatory factor of red R passages and blue channel B is carried out at white balance to background area image according to background area Reason.

7. the method that white balance according to claim 1 is processed, it is characterised in that described to be obtained in the human face region Reference zone, including：

In the human face region, position and extract tooth regions and define the tooth regions for reference zone.

8. the device that a kind of white balance is processed, it is characterised in that including：

Face recognition module, for recognizing human face region and the background area in addition to human face region；

Acquisition module, for obtaining reference zone in the human face region, wherein, the reference zone is the white in material object Region；

White balance processing module, for carrying out white balance treatment to whole image according to the reference zone；Or, according to described Reference zone the human face region is carried out white balance treatment and according to the background area to described in addition to human face region Region carries out white balance treatment.

9. the device that white balance according to claim 8 is processed, it is characterised in that the acquisition module includes：

Positioning unit, for eye areas or tooth regions in locating human face region；

Extraction unit, for extracting the white of the eye region in the eye areas or for extracting the tooth regions.

10. the device that white balance according to claim 8 is processed, it is characterised in that the acquisition module also includes：

Computing unit, the brightness of the brightness value acquisition eye areas of each pixel for calculating the eye areas is equal Value；

Unit is chosen, for choosing pixel of the eye areas brightness more than the luminance mean value, the set of the pixel of selection It is white of the eye region.

The device of 11. white balance treatment according to claim 9, it is characterised in that the processing module includes：

First statistic unit, for the RGB RGB color component average or maximum of statistical-reference region all pixels point；

First gain correction factor acquiring unit, for the RGB RGB color component average according to the reference zone or most Big value calculates the correction regulatory factor of red R passages and blue channel B；

First correction unit, for according to the correction regulatory factor of the red R passages of the reference zone and blue channel B to whole image Carry out white balance treatment.

The device of 12. white balance treatment according to claim 8, it is characterised in that the processing module also includes：

Second statistic unit, RGB RGB color component average or maximum for counting background area all pixels point；

Second gain correction factor acquiring unit, for according to the background area RGB RGB color component average or maximum Value calculates the correction regulatory factor of red R passages and blue channel B；

Second correction unit, for according to the correction regulatory factor of the red R passages in the background area and blue channel B to background area Image carries out white balance treatment.

A kind of 13. computer equipments, including memory, processor and the meter that store on a memory and can run on a processor Calculation machine program, following steps are realized during the computing device described program：

Recognition of face is carried out to image, human face region and the background area in addition to human face region is obtained；

Reference zone is obtained in the human face region, wherein, the light that reference zone described in material object reflexes to human eye has The ratio of blue, green, the red three kinds of coloured light in certain brightness and the light is identical；

White balance treatment is carried out to whole image according to the reference zone；Or, according to the reference zone to the face Region carries out white balance treatment；Background area is determined according to the human face region, the background area is except the human face region Region in addition；White balance treatment is carried out to the region in addition to human face region according to the background area.