CN115760816A - Method and device for determining illumination quality of face image and storage medium - Google Patents

Abstract

The invention discloses a method and a device for determining illumination quality of a face image and a storage medium. Wherein, the method comprises the following steps: analyzing a human face skin image from the human face image; acquiring a brightness coefficient and a brightness unevenness coefficient of a human face skin image; and determining the illumination quality of the face image according to the brightness coefficient and the brightness non-uniformity coefficient. The face illumination uniformity is determined by focusing only the face skin color area and showing a double peak phenomenon according to the uneven illumination, and finally face illumination quality evaluation is obtained by combining the face brightness determination, so that other image quality interference factors can be eliminated, and the image illumination brightness can be concentrated on the measurement.

Description

A method, device and storage medium for determining the illumination quality of a human face image

technical field

This article relates to image processing technology, especially a method, device and storage medium for determining the illumination quality of a face image.

Background technique

The lighting quality of the face image will affect the recognition results of the face image. When the lighting quality of the face image is poor, such as when the illumination of the face image is extremely bright, extremely dark or uneven, the accuracy of the face image recognition result will be reduced. Therefore, there is a need for an effective method for determining the quality of face illumination, which is conducive to guiding the composition of image data or optimizing the processing of image data, and is also conducive to analyzing the reasons for misrecognition and rejection of face recognition tasks.

The existing methods for determining the quality of face images mainly include: measuring the overall brightness based on the face image and its background image. In this method, the determination result of the light quality of the face image is greatly affected by the light quality of the background image; The image illumination is normal, but the background image illumination is extremely dark or extremely bright, which may reduce the determination result of the illumination quality of the face image; based on the priori that the feature difference of the left and right half face images is 0, the asymmetry of the left and right half face images caused by illumination The quality of the generated local variables is measured. This method is more strict with the prior. If the face is offset by a certain angle, then the left and right half-face images will no longer be symmetrical, and the resulting illumination quality evaluation indicators will no longer be accurate.

For the above problems, no effective solution has been proposed yet.

Contents of the invention

The present application provides a method, device and storage medium for determining the illumination quality of a human face image, which can more accurately determine the illumination quality of a human face image.

According to an aspect of an embodiment of the present invention, a method for determining the illumination quality of a human face image is provided, including: parsing a human face skin image from the human face image; acquiring the brightness coefficient and brightness unevenness of the human face skin image coefficient; determine the illumination quality of the human face image according to the brightness coefficient and the brightness non-uniformity coefficient.

Optionally, acquiring the brightness coefficient and brightness unevenness coefficient of the human face skin image includes: dividing the histogram curve of the human face skin image into bright areas and dark areas according to a preset pixel threshold; The histogram curve obtains the brightness coefficient and the brightness non-uniformity coefficient respectively, wherein the brightness non-uniformity coefficient is determined according to at least one of the following: a first coefficient used to represent the brightness difference between the bright area and the dark area , and a second coefficient used to represent the area difference between the bright area and the dark area.

Optionally, obtaining the brightness non-uniformity coefficient through the histogram curve includes: respectively determining a first pixel value and a second pixel value from the bright area and the dark area; combining the first pixel value , the second pixel value and the preset pixel threshold determine the first coefficient and the second coefficient respectively.

Optionally, determining the first pixel value and the second pixel value from the bright area and the dark area respectively includes: determining the pixel value that occurs most frequently in the bright area and the dark area, wherein the The pixel value corresponding to the peak value of the histogram curve in the bright area is the first pixel value, and the pixel value corresponding to the peak value of the histogram curve in the dark area is the second pixel value.

Optionally, determining the first coefficient in combination with the first pixel value, the second pixel value, and the preset pixel threshold includes: calculating a difference between the first pixel value and the pixel threshold , to obtain the first difference; calculate the difference between the second pixel value and the pixel threshold to obtain the second difference; and determine the first coefficient by calculating the average of the first difference and the second difference.

Optionally, determining the second coefficient in combination with the first pixel value, the second pixel value, and the preset pixel threshold includes: using the first pixel value, the second pixel value, and The preset pixel threshold value obtains the area of the bright area and the area of the dark area respectively; the ratio of the area of the bright area to the area of the dark area is used as the second coefficient.

Optionally, obtaining the area of the bright area and the area of the dark area includes: calculating the first pixel value, the second pixel value, the preset pixel threshold value and the pixel boundary value between two pairs The minimum value in the difference, using the minimum value as the peak width; taking the first pixel value as the central axis, the area of the histogram curve area included in the first left boundary and the first right boundary is the area of the bright area Area, wherein, the distances between the first left boundary and the first right boundary and the central axis are both the peak width; the second pixel value is the central axis, the second left boundary and the second right The area of the histogram curve area included in the boundary is the area of the dark area, wherein the distances between the second left boundary and the second right boundary and the central axis are both the peak width.

Optionally, obtaining the luminance coefficient through the histogram curve includes: calculating the mean and variance of the gray values of all pixels deviating from the central gray value based on the histogram curve; combining the mean and the The absolute value of the variance ratio is used as the brightness coefficient.

Optionally, the determining the illumination quality of the face image according to the brightness coefficient and the brightness non-uniformity coefficient includes: performing weighted calculation on the brightness coefficient, the first coefficient and the second coefficient and; determine the illumination quality of the human face image according to the result of the summation.

According to an aspect of an embodiment of the present invention, a device for determining the illumination quality of a human face image is provided, including: an analysis module for analyzing a human face skin image from a human face image; a processing module for obtaining the human face image The brightness coefficient and the brightness unevenness coefficient of the face skin image; an evaluation module, which is used to determine the illumination quality of the human face image according to the brightness coefficient and the brightness unevenness coefficient.

According to an aspect of an embodiment of the present invention, a computer-readable storage medium is provided, and the storage medium includes a stored program, wherein the program is the method for determining the illumination quality of a face image described in any one of the above items.

According to an aspect of an embodiment of the present invention, a processor is provided, and the processor is used to run a program, wherein, when the program is running, any method for determining the illumination quality of a human face image is executed.

In the embodiment of the present invention, by performing the following steps: analyzing the human face skin image from the human face image; obtaining the brightness coefficient and brightness unevenness coefficient of the human face skin image; The uniformity coefficient determines the lighting quality of the face image. By only focusing on the skin color area of the face, and according to the phenomenon of uneven illumination showing double peaks, the face illumination uniformity is evaluated, combined with the face brightness evaluation to finally obtain the face illumination quality evaluation, which is based on the brightness pixel distribution of the overall image. The measurement and calculation of relevant dimensions can not only be limited to regional semantic segmentation, but also measure the intensity of brightness contrast in uneven areas, reduce misjudgment caused by mild uneven skin color, and can well eliminate other image quality interference factors , focusing on the evaluation of image brightness.

Additional features and advantages of the application will be set forth in the description which follows, and, in part, will be obvious from the description, or may be learned by practice of the application. Other advantages of the present application can be realized and obtained through the schemes described in the specification and drawings.

Description of drawings

The accompanying drawings are used to provide an understanding of the technical solution of the present application, and constitute a part of the description, and are used together with the embodiments of the present application to explain the technical solution of the present application, and do not constitute a limitation to the technical solution of the present application.

Fig. 1 is the flow chart of the method for determining the illumination quality of a face image provided by the embodiment of the present application;

Fig. 2 is a schematic diagram of selecting human face skin images from multiple regional images provided by the embodiment of the present application;

FIG. 3 is a schematic diagram of a pixel histogram curve generated based on the image shown in FIG. 2 provided by the embodiment of the present application;

FIG. 4 is a schematic structural diagram of an apparatus for determining the illumination quality of a face image provided in an embodiment of the present application.

Detailed ways

The application describes a number of embodiments, but the description is illustrative rather than restrictive, and it will be obvious to those of ordinary skill in the art that within the scope of the embodiments described in the application, There are many more embodiments and implementations. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are possible. Except where expressly limited, any feature or element of any embodiment may be used in combination with, or substituted for, any other feature or element of any other embodiment.

This application includes and contemplates combinations of features and elements known to those of ordinary skill in the art. The disclosed embodiments, features and elements of this application can also be combined with any conventional features or elements to form unique inventive solutions as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive solutions to form yet another unique inventive solution as defined by the claims. It is therefore to be understood that any of the features shown and/or discussed in this application can be implemented alone or in any suitable combination. Accordingly, the embodiments are not to be limited except in accordance with the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

Furthermore, in describing representative embodiments, the specification may have presented a method and/or process as a particular sequence of steps. However, to the extent the method or process is not dependent on the specific order of steps described herein, the method or process should not be limited to the specific order of steps described. Other sequences of steps are also possible, as will be appreciated by those of ordinary skill in the art. Therefore, the specific order of the steps set forth in the specification should not be construed as limitations on the claims. In addition, claims for the method and/or process should not be limited to performing their steps in the order written, those skilled in the art can easily understand that these orders can be changed and still remain within the spirit and scope of the embodiments of the present application Inside.

The embodiment of the present application provides a method for determining the illumination quality of a face image, as shown in Figure 1, the method includes:

Step S101 parses out the face skin image from the face image;

The human face image may only include a human face; it may also include other parts of the human body (such as neck and shoulders) in addition to the human face, and may also include a background pattern;

Furthermore, in order to obtain more accurate light quality evaluation, this application can set the angle and blur parameter evaluation conditions of the face, can filter out high-quality face skin images, and perform subsequent analysis on the face images that reach the corresponding parameter setting range Process, so as to achieve subsequent more accurate light quality evaluation;

Step S102 obtains the brightness coefficient and the brightness unevenness coefficient of the above-mentioned human face skin image;

Step S103 determines the illumination quality of the human face image according to the brightness coefficient and the brightness non-uniformity coefficient.

The method for determining the illumination quality of a face image described in the embodiment of the present application only focuses on the skin area of the face, and eliminates the interference of the illumination quality of other parts of the image on the evaluation of the illumination quality of the face image. In addition, when determining the illumination quality of the face image, the application not only considers the strength of the illumination brightness, but also considers the uneven illumination brightness of the face caused by illumination, human body posture, etc., so that the finally determined illumination quality is more accurate. Specifically, this application should eliminate the interference to the light evaluation caused by some bright and dark areas that may only be connected to a small part of extremely strong light, and two large areas where the difference between bright and dark is not obvious.

In an exemplary embodiment, the method for parsing out a human face skin image from a human face image includes:

Decompose the face image based on the face parsing algorithm (such as the face parsing algorithm based on deep learning), and select the face skin image from it. After the face image is parsed, a label is assigned to each semantic component in the face area, and multiple areas include: hair, facial skin, eyes, nose, and mouth; as shown in Figure 2, from the multiple areas after parsing Extract human face skin images, including: nose area and facial skin area.

Remove the darker parts such as facial features and hair through face analysis, leaving only the skin part of the face for face lighting quality evaluation, eliminating the situation that affects the evaluation accuracy when there is background interference and other feature interference, such as in the brightness of the face It is normal but the background is extremely bright or extremely dark, which will lead to inaccurate evaluation results. This application only focuses on the skin color area of the face to obtain the most realistic light quality evaluation results of the actual area of the face.

In an exemplary embodiment, step S102 acquires the brightness coefficient and the brightness unevenness coefficient of the above-mentioned human face skin image, including:

Generate a histogram curve based on the above-mentioned human face skin image;

The brightness coefficient and the brightness non-uniformity coefficient are respectively obtained through the histogram curve.

In an exemplary embodiment, the above-mentioned brightness coefficient is obtained through the above-mentioned histogram curve, including:

Based on the above histogram curve, calculate the mean and variance of the gray value of all pixels from the central gray value;

The absolute value of the ratio of the above-mentioned mean value and the above-mentioned variance is used as the above-mentioned brightness coefficient.

Specifically, generate the histogram curve (also known as the grayscale histogram curve) of the human face skin image, the histogram curve generated based on the human face skin image shown in Figure 2 is as shown in Figure 3, wherein the histogram The abscissa of the figure represents the brightness in the image, gradually transitioning from all black to all white from left to right, and the range is [0,255]; the vertical axis represents the number of pixels in this brightness range in the image (also called frequency) , the overall brightness and darkness of the image can be obtained through the histogram curve, and the embodiment of the present application determines the brightness coefficient based on the histogram curve.

Further, as shown in FIG. 3 , the present application first calculates the mean value da of the gray scale deviation of the above-mentioned pixel histogram curve and the variance ma of the gray scale deviation; and then calculates the above-mentioned brightness coefficient according to the above-mentioned da and the above-mentioned ma.

In an exemplary embodiment, the method for calculating the grayscale deviation from the mean da and the variance ma of the grayscale deviation of the above-mentioned histogram curve includes:

Wherein, N represents the total number of pixels contained in the above-mentioned human face skin image; p _i represents the pixel value (also known as the gray value) of the ith pixel point; 128 represents the central gray value; j represents the pixel value, hist _j represents the frequency of occurrence of pixel value j.

The grayscale deviation from the mean da reflects the average brightness of the face skin image. If the average value is moderate, the visual effect is good; the grayscale deviation variance ma reflects the degree to which the pixel value deviates from the central grayscale value. The larger the value, the higher the grayscale level. The more scattered.

In an exemplary embodiment, calculating the above-mentioned brightness coefficient K according to the above-mentioned da and the above-mentioned ma includes:

K＝|da|/|ma| (3)

The smaller the brightness coefficient, the less the brightness value deviates from the central gray value. This application completes the brightness evaluation of the image based on the overall histogram curve, and measures the degree of dispersion of brightness data distribution through brightness coefficient statistics to measure the degree to which the brightness value deviates from the central gray value.

In an exemplary embodiment, obtaining the brightness non-uniformity coefficient through the above-mentioned histogram curve includes:

According to the preset pixel threshold, the histogram curve of the above-mentioned human face skin image is divided into a bright area and a dark area;

The brightness non-uniformity coefficient is obtained through the histogram curve, wherein the brightness non-uniformity coefficient is determined according to at least one of the following: a first coefficient used to represent the brightness difference between the bright area and the dark area, and a first coefficient used to represent the bright area The second coefficient of the difference from the area of the dark zone above.

Specifically, the above-mentioned bright area refers to the area where the pixel value of each pixel point in the histogram curve is greater than or equal to the preset pixel threshold; the above-mentioned dark area refers to the pixel value of each pixel point in the histogram curve. Areas with values less than a preset pixel threshold. The present application does not limit the method of determining the preset pixel threshold, such as the maximum entropy method, the minimum cross-entropy method, the maximum correlation method, the gray entropy method, and the maximum inter-class variance method, etc. may be used. Preferably, the maximum inter-class variance method can be used to find the pixel threshold T for distinguishing bright areas and dark areas; for dark areas, the pixel value of each pixel is less than T; for bright areas, the pixel value of each pixel is greater than or equal to T.

Specifically, this application divides the histogram curve corresponding to the human face skin image into bright areas and dark areas, instead of first dividing the human face skin image into blocks and then comparing the unevenness of the divided areas. It is not limited to image blocks, but accurately describes the unevenness of the brightness and darkness of the face based on the distribution of the overall pixels, avoiding the situation that the upper and lower parts of the yin and yang faces are not considered.

The present application realizes the judgment of unevenness through at least one of the following indicators: the first coefficient and the second coefficient. Among them, the first coefficient is used to indicate the brightness difference between the above-mentioned bright area and the above-mentioned dark area, that is, the contrast intensity of the uneven area illumination; the second coefficient is used to indicate the area difference between the above-mentioned bright area and the above-mentioned dark area, that is, the uneven area illumination corresponding ratio. Through these two indicators, it is not only possible to avoid misjudgment caused by light unevenness caused by slight uneven skin color, but also to accurately measure the brightness and contrast intensity of uneven areas without being limited by the division of image areas, thereby reducing the risk of blurred images. Misjudgment of uneven skin tone.

For the face skin image, the double peaks presented in the histogram curve are generally the normal skin color area of the face and the skin color area affected by the light, that is, the distribution of the two core pixels of the face skin color. Furthermore, this application uses the relationship between the double-peak peak pixel and the threshold distance and the area relationship of the double-peak area, rather than dividing the image into regions based on semantic information, or the priori of the symmetrical region of the face. Even if the facial illumination has asymmetry in various directions and the brightness of each illumination range is different, this application is still based on accurate statistics and description of the distribution of core pixels of human face skin color, and can objectively evaluate the unevenness of image illumination. Specifically, the brightness non-uniformity coefficient in this application includes two indexes, the first coefficient and the second coefficient, and both indexes depend on the peak pixel value and the preset pixel threshold.

In an exemplary embodiment, obtaining the above-mentioned brightness non-uniformity coefficient through the above-mentioned histogram curve includes:

determining a first pixel value and a second pixel value from said bright area and said dark area, respectively;

The first coefficient and the second coefficient are respectively determined in combination with the first pixel value, the second pixel value and the preset pixel threshold.

Specifically, before determining the brightness non-uniformity coefficient, the embodiment of the present application selects the first pixel value and the second pixel value respectively from the bright area and the dark area according to the same preset condition.

In an exemplary embodiment, the above-mentioned same preset condition includes: the pixel value that appears most frequently in the region. Since the abscissa of the pixel histogram represents the pixel value, and the ordinate represents the occurrence frequency of the corresponding pixel value, therefore, in an exemplary embodiment, the first pixel value and the second pixel value are respectively determined from the above-mentioned bright area and the above-mentioned dark area. Pixel value, including: determining the pixel value with the most frequent occurrence in the bright area and the dark area, wherein the pixel value corresponding to the peak value of the histogram curve in the bright area is the first pixel value, and the peak value of the histogram curve in the dark area corresponds to The pixel value of is the second pixel value.

As shown in Figure 3, due to uneven brightness, there will be peaks around the pixel threshold T of the histogram curve, and the maximum values pl and pr can be found on the left and right sides of the pixel threshold T, respectively, as the second pixel value and the first pixel value value. Specifically, in the bright area, search for the pixel histogram curve of the human face skin image to have a peak, and use the pixel value pr corresponding to the peak value as the first pixel value; in the dark area, search for the pixel histogram curve of the human face skin image When a peak appears, the pixel value pl corresponding to the peak value is used as the second pixel value.

The technical solution described in the embodiment of the present application is also applicable to the situation where there is no peak around the pixel threshold T (that is, there is no peak in the bright area and dark area). If there is no peak on the left or right of the pixel threshold T, pl=T or pr can be set correspondingly. =T. In addition, in the case of multiple sub-peaks in bright or dark areas, the sub-peaks caused by light are the real main peaks, and the remaining sub-peaks may be caused by small blemishes on the face or gradients on the face. And the two sub-peaks are not too far apart. If the heights of the secondary peaks are obviously different, that is, the difference between the peaks of the secondary peaks is greater than the difference threshold, then select the secondary peak with the highest peak, and use the pixel value corresponding to the peak as the pixel threshold; if the difference between the secondary peak heights is small, that is If the difference between the peaks of the sub-peaks is smaller than the difference threshold, the pixel value corresponding to any one of the sub-peaks is selected, or the final pixel value obtained by averaging the pixel values of all sub-peaks is determined as the pixel threshold.

Experiments have proved that the difference between the bright and dark areas of the image can be described by the pixel values corresponding to the peaks on the left and right sides of the pixel threshold, and the difference between the normal skin color area of the face and the skin color area of the same face affected by light can be accurately described. difference.

In an exemplary embodiment, determining the above-mentioned first coefficient in combination with the above-mentioned first pixel value, the above-mentioned second pixel value, and the above-mentioned preset pixel threshold value includes:

calculating the difference between the first pixel value and the pixel threshold to obtain a first difference;

calculating the difference between the second pixel value and the pixel threshold to obtain a second difference;

The first coefficient is determined by calculating an average of the first difference and the second difference.

For the face skin image, the double peaks presented in the histogram curve are generally the normal skin color area of the face and the skin color area affected by the light (or the skin color layered area such as the facial blush). Further, the present application can describe the contrast intensity of the two core pixels through the distance relationship between the bimodal peak pixel and the threshold value, and a smaller contrast intensity will reduce the final score, which can avoid problems caused by uneven skin color (blush on the face, etc.) The reason for the impact of the illumination uniformity description.

Specifically, according to the distance difference between the first pixel value and the second pixel value and the pixel threshold, the degree of difference between the pixel values of the bright area and the dark area, that is, the degree of illumination contrast, can be known. Taking Fig. 3 as an example, the difference between the above-mentioned first pixel value pr and the above-mentioned pixel threshold T is (pr-T), and the difference between the above-mentioned second pixel value pl and the above-mentioned pixel threshold T is (T-pl);

Then the first coefficient is:

This application uses the first coefficient to describe the average distance of the distance threshold of the double peaks, which represents a difference between the pixel value of the point with the largest number of pixel values in the darker illuminated area and the corresponding value in the brighter illuminated area, which can be more accurate In addition, it can also avoid the misjudgment of light unevenness caused by mild uneven skin color. For example, in areas where the difference between light and dark is not obvious such as facial blush, in the histogram curve, the Exists with a small shock amplitude. This application adopts the highest peak in the area. When the first coefficient is larger, the distance between the dark peak and the bright peak is larger, and the contrast intensity of the light is larger, the image is more uneven.

In addition, the embodiment of the present application also uses the ratio of the area of the bright area of the histogram curve to the area of the dark area as the second coefficient, which can accurately describe the proportion of core pixels of different skin colors of human faces, and avoid long-tail pixels from illuminating the face adverse effects on uniformity.

In an exemplary embodiment, determining the above-mentioned second coefficient in combination with the above-mentioned first pixel value, the above-mentioned second pixel value and the above-mentioned preset pixel threshold value includes:

Obtaining the area of the bright area and the area of the dark area respectively by using the first pixel value, the second pixel value, and the preset pixel threshold;

The ratio of the area of the bright area to the area of the dark area is used as the second coefficient.

The physical meaning corresponding to the second coefficient of the patent application is the proportion corresponding to the different lighting of the entire face. When evaluating the unevenness, it can reduce the impact of the brightness and darkness caused by connecting a small part of extremely strong light on the final unevenness.

Obtain the area of the above bright area and the area of the above dark area, including:

Calculating the minimum value of the above-mentioned first pixel value, the above-mentioned second pixel value, the difference between the above-mentioned preset pixel threshold value and the pixel boundary value, and using the above-mentioned minimum value as the peak width;

Taking the above-mentioned first pixel value as the central axis, the area of the histogram curve area contained in the first left boundary and the first right boundary is the area of the above-mentioned bright area, wherein, the above-mentioned first left boundary and the above-mentioned first right boundary are the same as the above-mentioned center Axis distances are the above peak widths;

Taking the above-mentioned second pixel value as the central axis, the area of the histogram curve area contained in the second left boundary and the second right boundary is the area of the above-mentioned dark area, wherein, the above-mentioned second left boundary and the above-mentioned second right boundary are the same as the above-mentioned middle Axis distances are the above-mentioned peak widths.

In addition, compared with using the full pixel segment of the area, in order to more accurately describe the pixels of the two core skin color states, long-tail pixels are excluded and added to the measurement index. Experiments show that the area ratio of the area formed by a small pixel distance around the peak can better describe The ratio of the current two core pixels.

Specifically, this application limits the peak range of bright and dark areas by introducing peak width, and calculates the difference between the first pixel value, the second pixel value, the preset pixel threshold and the pixel boundary value (including 0 and 255) The minimum value among the values, using the above minimum value as the peak width. Using the minimum value as the peak width can reduce the influence of long-tail pixels on the measurement index, and prevent the area from overlapping on the one hand. When selecting the area of the bright area, the area of the histogram curve area contained in the first left boundary and the first right boundary is the area of the bright area, where the first left boundary and the first right The distance between the boundary and the above-mentioned central axis is the peak width, and the same is true for the dark area.

Taking Figure 3 as an example, the area al of the dark area is:

The area ar of the bright area is:

Among them, halfw=min(|256-pr|, |pr-T|, |T-pl|, |Pl-0|), T is the pixel threshold; hist _i is the frequency of occurrence of pixel value i, which is The abscissa in the histogram curve is the ordinate corresponding to the pixel value i.

Correspondingly, the calculation method of the above second coefficient can be:

The second coefficient is the area ratio between the bright area and the dark area. The range of the second coefficient is greater than 1. When the coefficient is exactly 1, it means that the dark part and the bright part are exactly equal. The larger the second coefficient, the larger the area of a certain area , the more uniform the illumination of the image. By paying attention to the pixels whose peaks are two kinds of skin color states, the exclusion part may only be bright and dark caused by a small part of extremely strong light, thereby reducing the misjudgment of image unevenness.

In an exemplary embodiment, determining the illumination quality of the above-mentioned face image according to the above-mentioned brightness coefficient and the above-mentioned brightness non-uniformity coefficient includes:

performing weighted summation of the brightness coefficient, the first coefficient and the second coefficient;

The illumination quality of the above-mentioned face image is determined according to the result of the summation.

Perform weighted summation of the brightness coefficient, the first coefficient and the second coefficient according to weights, such as score=w ₁ *K+w ₂ *dis+w ₃ *ratio, wherein w ₁ , w ₂ , and w ₃ represent The weights of the three coefficients can be adaptively set according to experiments or requirements; the illumination quality of the above-mentioned face image is determined according to the result score of the summation, the larger the score value, the better the illumination quality; conversely, the smaller the score value, the better the illumination quality. worse.

This application only focuses on the skin color area of the face, without background interference to evaluate the lighting quality of the face, and evaluates the uniformity of the lighting on the face according to the phenomenon of uneven lighting showing double peaks, combined with the evaluation of the brightness of the face, finally obtains the evaluation of the lighting quality of the face , this evaluation does not rely on semantic region segmentation, but is based on the brightness pixel distribution of the overall image to perform measurement calculations of relevant dimensions, which is not limited to the assumption of symmetry between the left and right face parts, and can also measure the brightness contrast of uneven regions The level of intensity reduces the misjudgment caused by mild facial uneven skin tone, which can well eliminate other image quality interference factors and focus on the evaluation of image brightness.

The embodiment of the present application provides a device for determining the illumination quality of a human face image. As shown in FIG. Obtaining the brightness coefficient and brightness unevenness coefficient of the above-mentioned human face skin image; the evaluation module 430 is configured to determine the illumination quality of the above-mentioned human face image according to the above-mentioned brightness coefficient and the above-mentioned brightness unevenness coefficient.

The device for determining the illumination quality of a face image described in the embodiment of the present application only focuses on the skin area of the face, and eliminates the interference of the illumination quality of other parts of the image on the evaluation of the illumination quality of the face image. In addition, when determining the illumination quality of the face image, the application not only considers the strength of the illumination brightness, but also considers the uneven illumination brightness of the face caused by illumination, human body posture, etc., so that the finally determined illumination quality is more accurate. Specifically, this application excludes the interference to the lighting evaluation caused by the bright and dark areas caused by the connection of a small part of extremely strong light, and the two large areas where the difference between bright and dark is not obvious.

In an exemplary embodiment, the above processing module 410 includes:

The first processing module is used to divide the histogram curve of the above-mentioned human face skin image into bright areas and dark areas according to a preset pixel threshold;

The second processing module is configured to respectively obtain the above-mentioned brightness coefficient and the above-mentioned brightness non-uniformity coefficient through the above-mentioned histogram curve, wherein the above-mentioned brightness non-uniformity coefficient is determined according to at least one of the following: used to represent the brightness difference between the above-mentioned bright area and the above-mentioned dark area The first coefficient, and the second coefficient used to represent the area difference between the above-mentioned bright area and the above-mentioned dark area.

Specifically, the above-mentioned bright area refers to the area where the pixel value of each pixel point in the histogram curve is greater than or equal to the preset pixel threshold; the above-mentioned dark area refers to the pixel value of each pixel point in the histogram curve. Areas with values less than a preset pixel threshold. The present application does not limit the method for determining the preset pixel threshold, such as the maximum entropy method, the minimum cross-entropy method, the maximum correlation method, the gray entropy method, and the maximum inter-class variance method. Preferably, the maximum inter-class variance method can be used to find the pixel threshold T for distinguishing bright areas and dark areas; for dark areas, the pixel value of each pixel is less than T; for bright areas, the pixel value of each pixel is greater than or equal to T.

Specifically, this application divides the histogram curve corresponding to the human face skin image into bright areas and dark areas, instead of first dividing the human face skin image into blocks and then comparing the unevenness of the divided areas. Limited to image blocks, this application is based on the distribution of the overall pixels to accurately describe the unevenness of the brightness and darkness of the human face, avoiding the situation that the upper and lower parts of the yin and yang faces are not considered.

In an exemplary embodiment, the above-mentioned second processing module includes a brightness coefficient unit, including:

The first processing unit is configured to calculate, based on the above-mentioned histogram curve, the mean and variance of the gray values of all pixels deviating from the central gray value;

The second processing unit is configured to use the absolute value of the ratio of the above-mentioned mean value and the above-mentioned variance as the above-mentioned brightness coefficient.

The smaller the brightness coefficient, the less the brightness value deviates from the central gray value. This application completes the brightness evaluation of the image based on the overall histogram curve, and measures the degree of dispersion of brightness data distribution through brightness coefficient statistics to measure the degree to which the brightness value deviates from the central gray value.

In an exemplary embodiment, the above-mentioned second processing module also includes a non-uniform coefficient unit, including:

a third processing unit, configured to determine a first pixel value and a second pixel value from the bright area and the dark area, respectively;

The fourth processing unit is configured to combine the first pixel value, the second pixel value and the preset pixel threshold to determine the first coefficient and the second coefficient respectively.

The present application realizes the judgment of unevenness through at least one of the following indicators: the first coefficient and the second coefficient. Among them, the first coefficient is used to represent the brightness difference between the above-mentioned bright area and the above-mentioned dark area, that is, the contrast intensity of the illumination in the non-uniformity area; the second coefficient is used to represent the area difference between the above-mentioned bright area and the above-mentioned dark area, that is, the non-uniformity The proportion corresponding to the illumination in the unevenness area, through these two indicators, not only can avoid the misjudgment caused by uneven skin color to uneven light, but also measure the brightness contrast intensity of the uneven area on the basis of not being limited to the division of image areas to reduce the misjudgment of uneven skin tone on the face.

In an exemplary embodiment, the uneven coefficient processing unit further includes a first uneven coefficient subunit, including:

A first processing subunit, configured to calculate a difference between the first pixel value and the pixel threshold to obtain a first difference;

a second processing subunit, configured to calculate a difference between the second pixel value and the pixel threshold to obtain a second difference;

A third processing subunit, configured to determine the first coefficient by calculating an average of the first difference and the second difference.

This application uses the first coefficient to describe the average distance of the distance threshold of the double peaks, which represents a difference between the pixel value of the point with the largest number of pixel values in the darker illuminated area and the corresponding value in the brighter illuminated area, which can be more accurate In addition, it can also avoid the misjudgment of light unevenness caused by mild uneven skin color. For example, in areas where the difference between light and dark is not obvious such as facial blush, in the histogram curve, the Exists with a small shock amplitude. This application adopts the highest peak in the area. When the first coefficient is larger, the distance between the dark peak and the bright peak is larger, and the contrast intensity of the light is larger, the image is more uneven.

In addition, the embodiment of the present application also uses the ratio of the area of the bright area of the histogram curve to the area of the dark area as the second coefficient, which can accurately describe the proportion of core pixels of different skin colors of human faces, and avoid long-tail pixels from illuminating the face adverse effects on uniformity.

In an exemplary embodiment, the uneven coefficient unit further includes a second uneven coefficient subunit, including:

A fourth processing subunit, configured to respectively obtain the area of the bright area and the area of the dark area through the first pixel value, the second pixel value, and the preset pixel threshold;

The fifth processing subunit is configured to use the ratio of the area of the bright area to the area of the dark area as the second coefficient.

The physical meaning corresponding to the second coefficient of the patent application is the proportion corresponding to the different lighting of the entire face. When evaluating the unevenness, it can reduce the impact of the brightness and darkness caused by connecting a small part of extremely strong light on the final unevenness. The second coefficient is the area ratio between the bright area and the dark area. The range of the second coefficient is greater than 1. When the coefficient is exactly 1, it means that the dark part and the bright part are exactly equal. The larger the second coefficient, the larger the area of a certain area , the more uniform the illumination of the image.

This application only focuses on the skin color area of the face, without background interference to evaluate the lighting quality of the face, and evaluates the uniformity of the lighting on the face according to the phenomenon of uneven lighting showing double peaks, combined with the evaluation of the brightness of the face, finally obtains the evaluation of the lighting quality of the face , this evaluation does not rely on semantic region segmentation, but is based on the brightness pixel distribution of the overall image to perform measurement calculations of relevant dimensions, which is not limited to the assumption of symmetry between the left and right face parts, and can also measure the brightness contrast of uneven regions The level of intensity reduces the misjudgment caused by mild facial uneven skin tone, which can well eliminate other image quality interference factors and focus on the evaluation of image brightness.

The embodiment of the present application also provides a computer-readable storage medium, where one or more programs are stored in the computer-readable storage medium, and the one or more programs can be executed by one or more processors to implement the following: The method described in any of the previous examples.

According to another aspect of the embodiments of the present invention, a processor is also provided, and the processor is used to run a program; wherein, when the program is running, the method described in any one of the above-mentioned embodiments is executed.

The serial numbers of the above embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

In the above-mentioned embodiments of the present invention, the descriptions of each embodiment have their own emphases, and for parts not described in detail in a certain embodiment, reference may be made to relevant descriptions of other embodiments.

Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, the functional modules/units in the system, and the device can be implemented as software, firmware, hardware, and an appropriate combination thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be composed of several physical components. Components cooperate to execute. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those of ordinary skill in the art, the term computer storage media includes both volatile and nonvolatile media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data. permanent, removable and non-removable media. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cartridges, tape, magnetic disk storage or other magnetic storage devices, or can Any other medium used to store desired information and which can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media .

Claims (12)

1. A method for determining the quality of illumination of a human face image, comprising: Parse the face skin image from the face image; Obtain the brightness coefficient and brightness unevenness coefficient of the human face skin image; Determine the illumination quality of the human face image according to the brightness coefficient and the brightness non-uniformity coefficient. 2. The method of claim 1, wherein, Obtaining the brightness coefficient and brightness unevenness coefficient of the human face skin image, including: According to the preset pixel threshold, the histogram curve of the human face skin image is divided into bright areas and dark areas; Obtaining the brightness coefficient and the brightness non-uniformity coefficient respectively through the histogram curve, wherein the brightness non-uniformity coefficient is determined according to at least one of the following: A first coefficient used to represent the brightness difference between the bright area and the dark area, and a second coefficient used to represent the area difference between the bright area and the dark area. 3. The method of claim 2, wherein, Obtaining the brightness non-uniformity coefficient through the histogram curve includes: determining a first pixel value and a second pixel value from said light region and said dark region, respectively; The first coefficient and the second coefficient are respectively determined in combination with the first pixel value, the second pixel value and the preset pixel threshold. 4. The method of claim 3, wherein, Determining a first pixel value and a second pixel value from the bright area and the dark area, respectively, comprising: Determining the pixel value with the highest frequency of occurrence in the bright area and the dark area, wherein the pixel value corresponding to the peak value of the histogram curve in the bright area is the first pixel value, and the peak value of the histogram curve in the dark area is The corresponding pixel value is the second pixel value. 5. The method of claim 3, wherein, Determining the first coefficient in combination with the first pixel value, the second pixel value and the preset pixel threshold includes: calculating the difference between the first pixel value and the pixel threshold to obtain a first difference; calculating the difference between the second pixel value and the pixel threshold to obtain a second difference; The first coefficient is determined by calculating an average of the first difference and the second difference. 6. The method of claim 3, wherein, Determining the second coefficient in combination with the first pixel value, the second pixel value and the preset pixel threshold includes: Obtaining the area of the bright region and the area of the dark region respectively by the first pixel value, the second pixel value and the preset pixel threshold; The ratio of the area of the bright area to the area of the dark area is used as the second coefficient. 7. The method of claim 6, wherein, Acquiring the area of the bright area and the area of the dark area, including: calculating the minimum value of the first pixel value, the second pixel value, the difference between the preset pixel threshold value and the pixel boundary value, and using the minimum value as the peak width; Taking the first pixel value as the central axis, the area of the histogram curve area contained by the first left boundary and the first right boundary is the area of the bright area, wherein the first left boundary and the first right The distance between the boundary and the central axis is the peak width; Taking the second pixel value as the central axis, the area of the histogram curve area contained by the second left boundary and the second right boundary is the area of the dark area, wherein the second left boundary and the second right The distance between the boundary and the central axis is the peak width. 8. The method of claim 2, wherein, Obtaining the brightness coefficient through the histogram curve includes: Based on the histogram curve, calculate the mean and variance of the gray values of all pixels deviating from the central gray value; The absolute value of the ratio of the mean value to the variance is used as the brightness coefficient. 9. The method according to any one of claims 2-8, characterized in that, The determining the illumination quality of the face image according to the brightness coefficient and the brightness non-uniformity coefficient includes: performing weighted summation of the brightness coefficient, the first coefficient and the second coefficient; Determine the illumination quality of the face image according to the summation result. 10. A device for determining the illumination quality of a human face image, comprising: The analysis module is used to analyze the human face skin image from the human face image; A processing module, configured to obtain a brightness coefficient and a brightness unevenness coefficient of the human face skin image; An evaluation module, configured to determine the illumination quality of the human face image according to the brightness coefficient and the brightness non-uniformity coefficient. 11. A computer-readable storage medium, characterized in that the storage medium includes a stored program, wherein the program executes the method for determining the illumination quality of a face image according to any one of claims 1 to 9. 12. A processor, characterized in that the processor is used to run a program, wherein the method for determining the illumination quality of a human face image according to any one of claims 1 to 9 is executed when the program is running.

Images