A color correction method for ultra-high-definition cameras

Abstract

This application relates to the field of image color enhancement technology, specifically to a color correction method for an ultra-high-definition camera. The method includes: acquiring each frame of video data; dividing each frame of image into multiple regions on an average basis; obtaining the target presence coefficient of each region in each frame of image based on the color difference between different color blocks in each region, the proportion of edge pixels, and the dispersion of the average grayscale value of all edges; and obtaining the correction coefficient of each region in each frame of image by combining the total number of edges in the neighboring regions of each region in each frame of image, the color difference and texture difference in all adjacent frames; further determining whether each region of each frame of image needs to be segmented into smaller regions; when segmentation is required, obtaining the side length of the segmented smaller regions in each frame of image; and finally obtaining the color-corrected video. This application improves the color correction effect of video by dividing the region where the target object is located in each frame of image into smaller regions.

Description

Technical Field

[0001] This application relates to the field of image color enhancement technology, and more specifically to a color correction method for ultra-high-definition cameras. Background Technology

[0002] In the field of photogrammetry, imaging technology has developed rapidly. Multi-camera systems, such as panoramic cameras, free-viewpoint cameras, high-speed video shooting, and light field imaging, have many popular applications in people's daily lives, such as image-based 3D reconstruction, augmented reality, and autonomous driving. However, during image stitching, due to differences in camera settings, lighting conditions, and viewing angles, there is usually a certain color difference between adjacent images. The reproduction effect varies greatly between different hardware devices. Therefore, color correction technology is usually required to correct video images.

[0003] Existing color correction methods typically focus on processing single-frame images, lacking effective consideration for the temporal consistency of video. Due to the complex changes in lighting and color between video frames, they often fail to adequately guarantee color consistency between consecutive frames, easily leading to unsmooth visual effects in the video, and even causing jumps and flickering. Furthermore, most existing color correction techniques rely on overall style mapping and correction of single-frame images. However, in dynamic video images, the color representation of target areas often differs from that of background areas. Therefore, global adjustments may lead to color distortion in local areas, resulting in significant color deviations in the video image. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a color correction method for ultra-high-definition cameras to solve the existing issues.

[0005] The color correction method for an ultra-high-definition camera proposed in this application adopts the following technical solution:

[0006] One embodiment of this application provides a color correction method for an ultra-high-definition camera, the method comprising the following steps:

[0007] Acquire each frame of the video data to be color corrected;

[0008] Each frame of the image is divided into multiple regions on an average basis. All pixels in each region are clustered according to the pixel value of each pixel in each region. The color block difference of each region in each frame of the image is obtained by the minimum difference between the average pixel values ​​of each adjacent cluster in each region. The target existence coefficient of each region in each frame of the image is obtained by combining the proportion of edge pixels in each region in each frame of the image and the dispersion of the average gray value of all edges in each region.

[0009] Based on the total number of edges of all neighboring regions corresponding to each region in each frame of the current video, and the difference in pixel values ​​of each region in all adjacent frames of the current video, the target region weight of each region is obtained, and combined with the difference in LBP values ​​of each region in all adjacent frames of the current video, the multi-frame target factor of each region is obtained.

[0010] Based on the target presence coefficient of each region in each frame image and the target factor of multiple frames, the correction coefficient of each region in each frame image is obtained. Then, it is determined whether each region in each frame image needs to be segmented into smaller regions. When segmentation is required, the side length of the segmented smaller regions in each frame image is obtained, and then the color-corrected video is obtained.

[0011] Preferably, the specific process of clustering all pixels in each region is as follows: the average value of the pixel values ​​of the R, G, and B channels of each pixel is recorded as the color value of each pixel; the color values ​​of all pixels in each region are used as the input of the DPC density clustering algorithm, and the absolute difference between the color values ​​of pixels is used as the distance metric, and multiple clusters are output.

[0012] Preferably, the method for obtaining the color block difference degree of each region in each frame image is as follows: the minimum value of the difference between the mean color values ​​of pixels in all two adjacent clusters in each region of each frame image is calculated, and this value is used as the color block difference degree of each region in each frame image.

[0013] Preferably, the formula for calculating the target presence coefficient in each region of each frame image is as follows: In the formula, The coefficient representing the presence of the target in the i-th region of the k-th frame image. Let be the edge density of the i-th region in the k-th frame image, where the ratio of the total number of edge pixels to the total number of pixels in each region is used as the edge density of each region. Let be the variance of the mean grayscale values ​​of all edges in the i-th region of the k-th frame image. Let i be the color patch difference in the i-th region of the k-th frame image. This is a preset constant.

[0014] Preferably, the formula for calculating the target region weight of each region is as follows: In the formula, Let i be the target region weight for the i-th region. This represents the total number of edges of all neighboring regions of the i-th region in the u-th frame of the current video. , are the average color values ​​of the i-th region in the u-th and (u-1)-th frames of the current video, respectively, where U is the total number of frames in the current video.

[0015] Preferably, the method for obtaining the multi-frame target factors of each region is as follows: the cumulative result of the difference of LBP values ​​in the same region in all adjacent frames of the current video is calculated, and the multi-frame target factors of the same region are positively correlated with the target region weight of the same region and negatively correlated with the cumulative result.

[0016] Preferably, the correction coefficient of each region in each frame image is the normalized result of the product of the target existence coefficient of each region and the target factor of multiple frames.

[0017] Preferably, the specific process of determining whether each region of each frame image needs to be divided into smaller regions is as follows: when the correction coefficient of any region of each frame image is greater than or equal to a preset segmentation threshold, it is determined that the region needs to be divided into smaller regions; otherwise, it is determined that the region does not need to be divided into smaller regions.

[0018] Preferably, the formula for calculating the side length of the small regions after segmentation in each frame image is as follows: In the formula, Let v be the side length of the sub-region after segmenting the i-th region in the k-th frame image, and v be the preset initial side length. Let be the correction coefficient for the i-th region in the k-th frame image, and e be a preset coefficient. , This is the floor function.

[0019] Preferably, the specific process of obtaining the color-corrected video is as follows: each region is divided into small regions according to the side length of the corresponding small regions; each frame of the video data to be color-corrected uses the undivided regions and the divided small regions as inputs, and uses a histogram equalization algorithm to perform color correction on each region or small region, outputting each region or small region after color correction, and merging all the color-corrected regions and small regions in each frame to obtain a video in which each frame has been color-corrected.

[0020] This application has at least the following beneficial effects:

[0021] This application addresses the problem that traditional color correction algorithms often only correct the overall color of the video image, ignoring the differences between the target and background regions, as well as the continuity between video frames. This leads to local distortion and poor color consistency between consecutive frames during color correction. By analyzing the image features of the area where the subject is located in each frame of a video captured by an ultra-high-definition camera, a target presence coefficient is constructed, which can initially distinguish the target and background regions in each frame. By analyzing the color and texture changes of the background and target regions between consecutive frames in the video, a correction coefficient is constructed, which can more accurately distinguish the area where the subject is located in each frame of the video. This allows each region to be divided into smaller regions, and then a histogram equalization algorithm is used to perform color correction on each region, thereby improving the color correction effect of the video. Attached Figure Description

[0022] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating the steps of a color correction method for an ultra-high-definition camera provided in this application;

[0024] Figure 2 A flowchart illustrating the process of obtaining the correction coefficients for each region in each frame of the image provided in this application;

[0025] Figure 3 The vector waveform diagram before color correction provided for this application;

[0026] Figure 4 The color-corrected vector waveform provided for this application. Detailed Implementation

[0027] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a color correction method for an ultra-high-definition camera proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0029] The following description, in conjunction with the accompanying drawings, details a specific scheme for a color correction method for an ultra-high-definition camera provided in this application.

[0030] This application provides a color correction method for an ultra-high-definition camera in one embodiment. Specifically, it provides the following color correction method for an ultra-high-definition camera. Please refer to [link to relevant documentation]. Figure 1 The method includes the following steps:

[0031] Step 1: Obtain each frame of the video data to be color corrected.

[0032] The required video data is captured by an ultra-high-definition camera and uploaded to the video image correction system. The video data is marked as video data to be color corrected. The grayscale image is converted by the grayscale value averaging method for each captured image, and finally the RGB image and grayscale image of each frame are obtained.

[0033] Step 2: Divide each frame image into multiple regions on an average basis. Cluster all pixels in each region according to the pixel value of each pixel in each region. Obtain the color block difference degree of each region in each frame image based on the minimum difference between the average pixel values ​​of each adjacent cluster in each region. Combine the proportion of edge pixels in each region in each frame image with the dispersion of the average gray value of all edges in each region to obtain the target existence coefficient of each region in each frame image.

[0034] For video image data, the shooting has a specific purpose, namely, a clear subject, such as a person or scene. Therefore, during the shooting process, the camera focuses on the corresponding subject, making the image features of the subject (i.e., the target area) clearer and more prominent, while simultaneously making non-subject objects (i.e., the background) more blurred. As a result, in the overall video image, the overall color zoning of the subject's location is clear, and the stripe edges of the subject are more defined. Conversely, colors in the background area, due to blurring, exhibit blurred edges, significant brightness differences between edges, and relatively lower edge density, with stripe edges being even more blurred. Specifically, if a certain area in the video image is the target area, the edges of different color blocks within it are clear, with a high degree of transition and distinct stripes, while in the background area, the brightness differences between edges are significant, and the distribution density is lower.

[0035] Taking a single frame of video image data to be color corrected as an example, the image is divided into regions with a side length of v×v, where v is a preset initial side length. In this embodiment, v is set to 30, but the implementer can adjust it according to the actual accuracy requirements. The average value of the R, G, and B channels of each pixel is recorded as the color value of each pixel. Taking the i-th region in the k-th frame as an example, the color values ​​of all pixels in the i-th region are used as the input of the DPC density clustering algorithm, with a cutoff distance of 3. The absolute difference between the color values ​​of pixels is used as the distance metric, and the output is multiple clusters, each cluster being a color patch in the i-th region. It should be noted that the DPC density clustering algorithm is a well-known technology, and the specific process will not be described in detail.

[0036] Furthermore, to characterize the edge stripe distribution in a single region, the i-th region is used as the input to the edge detection algorithm to obtain the total number of edges in the i-th region. The ratio of the total number of pixels on the edges in the i-th region to the total number of pixels in the region is used as the edge density of the i-th region.

[0037] In a preferred embodiment, the color patch difference degree of each region in each frame image is obtained based on the minimum difference between the average pixel values ​​of adjacent clusters in each region. In this embodiment, the color patch difference degree of the i-th region in the k-th frame image is denoted as... Specifically, the minimum difference between the mean color values ​​of pixels in all two adjacent clusters in each region of each frame image is used as the color patch difference degree of each region in each frame image.

[0038] As a preferred implementation, the target presence coefficient of each region in each frame image is obtained based on the color block difference of each region in each frame image, the proportion of edge pixels, and the dispersion of the average gray value of all edges. This coefficient is used to characterize the probability that each region in each frame image belongs to the area where the video shooting object is located.

[0039] In this embodiment, the target presence coefficient of the i-th region in the k-th frame image is denoted as... Its specific expression is: In the formula, The coefficient representing the presence of the target in the i-th region of the k-th frame image. Let be the edge density in the i-th region of the k-th frame image. Let be the variance of the mean grayscale values ​​of all edges in the i-th region of the k-th frame image. Let i be the color patch difference in the i-th region of the k-th frame image. This is a preset constant used to prevent the denominator from being 0. In this embodiment... .

[0040] The meaning of this relationship is: the more details in the i-th region of the k-th frame image, the higher the degree of transition between the overall color blocks, the greater the edge density in the region, the smaller the brightness difference between the edges, that is, the smaller the variance of the mean gray value of all edges, and the greater the overall difference of each color block in the region, indicating that the region is more likely to be the area where the shooting object exists in the video image, and the region size should be reduced, so as to perform more detailed color correction on the region and prevent color distortion in the region.

[0041] Step 3: Based on the total number of edges of all neighboring regions corresponding to each region in each frame of the current video, and the difference in pixel values ​​of each region in all adjacent frames of the current video, obtain the target region weight of each region, and combine the difference in LBP values ​​of each region in all adjacent frames of the current video to obtain the multi-frame target factor of each region.

[0042] During color correction, because video shooting is continuous, the subject in the video image may differ at different shooting times. Furthermore, during continuous shooting, the lighting environment faced by the subject may change rapidly, and the focus may change due to shooting movement, resulting in blurring of local areas of the main subject in the image. If only the calculation is performed using the method in step two, the area of ​​the subject will be treated as the background area, leading to color distortion in the target area during color correction. Therefore, further analysis is needed.

[0043] Specifically, as video recording progresses, the background may continuously change, but the main subject remains within the video images. That is, as multiple frames of video are captured, the difference in the target area where the subject is located between frames remains relatively small, while the background area changes more significantly with the movement of the video. Specifically, if a single area is within the target area of ​​the subject across multiple frames, the number of edge stripes is greater, and the color fluctuation of that single area between frames is weaker. Furthermore, even if the lighting environment changes, the overall color change of the target area where the subject is located is relatively small, and the overall texture feature differences are small. The areas within the eight-neighborhood of each region are denoted as its neighboring regions.

[0044] Based on the above analysis, the target region weight of each region is obtained according to the total number of edges of all neighboring regions corresponding to each region in each frame of the current video, and the difference in pixel values ​​of each region in all adjacent frames of the current video.

[0045] In this embodiment, the target region weight of the i-th region is denoted as... Its specific expression is: In the formula, Let i be the target region weight for the i-th region. This represents the total number of edges of all neighboring regions of the i-th region in the u-th frame of the current video. , , where are the average color values ​​of the i-th region in the u-th and (u-1)-th frames of the current video, respectively, and U is the total number of frames in the current video. The larger the target region weight value of the i-th region, the more edges the region's neighboring regions have in the multiple frames, and the smaller the overall color difference of the region in the multiple frames, making the region more likely to be the target region.

[0046] Based on the target region weights of each region and the differences in LBP values ​​of each region across all adjacent frames of the current video, multi-frame target factors are obtained for each region to characterize the probability that each region is the location of the subject being filmed. The process of obtaining the multi-frame target factors involves: summing the differences in LBP values ​​of the same region across all adjacent frames of the current video. The multi-frame target factors for the same region are positively correlated with the target region weights of that region and negatively correlated with the summed values. It should be noted that the positive correlation means the dependent variable increases (decreases) as the independent variable increases (decreases), and the negative correlation means the dependent variable decreases (increases) as the independent variable increases (decreases). The specific calculation relationship is not specifically limited in this application. Preferably, in this embodiment, the multi-frame target factor of the i-th region is denoted as... Its specific expression is: In the formula, For the multi-frame target factor of the i-th region, Let i be the target region weight for the i-th region. , , respectively, are the decimal values ​​of the LBP values ​​of the i-th region in the u-th and (u-1)-th frames of the current video, where U is the total number of frames in the current video. This is a preset constant. It should be noted that the calculation of LBP values ​​for a single region is a well-known technique, and the specific process will not be described in detail here.

[0047] The meaning of this relationship is: between multiple frames of a video, if... The larger the value, the more edges the neighboring regions of the i-th region have in multiple frames, the smaller the overall color difference of the region in multiple frames, and the smaller the overall texture feature difference of the region in multiple frames. This indicates that the region is more likely to be the target area of ​​the shooting object during video dynamic shooting, and the more detailed the color correction of the region should be.

[0048] Step 4: Obtain the correction coefficients of each region in each frame image based on the target presence coefficients of each region in each frame image and the target factors of multiple frames. Then determine whether each region in each frame image needs to be segmented into smaller regions. When segmentation is required, obtain the side length of the smaller regions after segmentation in each frame image, and then obtain the color-corrected video.

[0049] Furthermore, the correction coefficients for each region in each frame are obtained based on the target presence coefficients of each region in each frame and the target factors of multiple frames. In this embodiment, the normalized result of the product of the target presence coefficient of the i-th region in the k-th frame and the target factors of multiple frames is used as the correction coefficient of the i-th region in the k-th frame.

[0050] The meaning of this relationship is: in the k-th frame of the current video, the greater the probability that a target exists in the i-th region, and the greater the probability that a target exists in this region in multiple frames, the more likely it is that the video is capturing an object in this region. Therefore, when performing color correction on the video image, more detailed color correction should be performed on this region to prevent color distortion in this region and ensure the color accuracy of this region.

[0051] Similarly, the correction coefficients for each region in each frame of the current video can be calculated. The process for obtaining the correction coefficients for each region in each frame is as follows: Figure 2 As shown.

[0052] Furthermore, the correction coefficients of all regions in all frames of the current video are used as input, and cross-validation is used to output a segmentation threshold, which is recorded as the preset segmentation threshold. When the correction coefficient of a single region within a single frame of the video is greater than or equal to the preset segmentation threshold, it is determined that the region contains the target object captured in this video. During the correction process, this region needs to be segmented into smaller regions for more detailed color correction; otherwise, it is determined that the region is background content in the current video, and this region does not need to be segmented into smaller regions.

[0053] Based on the above analysis, when the correction coefficient of the i-th region in the k-th frame image is greater than or equal to the preset segmentation threshold, the side length of the segmented small region in that frame image is calculated: In the formula, Let v be the side length of the sub-region after segmenting the i-th region in the k-th frame image, and v be the preset initial side length. Let be the correction coefficient for the i-th region in the k-th frame image, and e be a preset coefficient. In this embodiment, e is 4. This is a rounding function used to round the input data to the nearest integer. It should be noted that if the i-th region cannot be divided by the side length... When divisible, that is, after dividing into small regions, a portion of the length that is less than a certain value remains at the edges. When dealing with a region, the remaining edge portion is treated as a separate small region.

[0054] Using the above method, each region can be divided into smaller regions based on the side length of the corresponding smaller regions. Further, the undivided regions and the divided smaller regions in each frame of the video data to be color-corrected are taken as input, and a histogram equalization algorithm is used to perform color correction on each region or smaller region, outputting each region or smaller region after color correction. All color-corrected regions and smaller regions in each frame are then merged to obtain a video where each frame has undergone color correction.

[0055] Finally, a vector oscilloscope is used to measure the hue and color distribution in each frame of the image to determine if any color difference still exists. A vector oscilloscope is a comprehensive oscilloscope used to measure the hue and saturation of an image. By observing the vector oscilloscope extended at different angles, the hue of the image can be obtained. By determining the center position of the vector waveform relative to the crosshairs, a color imbalance in the image can be quickly detected. If the vector waveform deviates from the center of the crosshairs, it indicates a color imbalance; if the vector waveform coincides with the crosshairs, it indicates color balance. Figure 3 The vector waveform diagram before color correction provided in this application Figure 4 The color-corrected vector waveform provided for this application; through Figure 3 and Figure 4 The comparison shows that after color correction of the image using the method provided in this application, the overall color tone of the image is more uniform, and the image does not show any bias towards any particular color, indicating a good correction effect.

[0056] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0057] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0058] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A color correction method for an ultra-high definition video camera, characterized by, The method includes the following steps: Acquire each frame of the video data to be color corrected; Each frame of the image is divided into multiple regions on an average basis. All pixels in each region are clustered according to the pixel value of each pixel in each region. The color block difference degree of each region in each frame of the image is obtained by the minimum difference between the average pixel values ​​of each adjacent cluster in each region. The target existence coefficient of each region in each frame of the image is obtained by combining the proportion of edge pixels in each region in each frame of the image and the dispersion of the average gray value of all edges in each region. Based on the total number of edges of all neighboring regions corresponding to each region in each frame of the current video, and the difference in pixel values ​​of each region in all adjacent frames of the current video, the target region weight of each region is obtained, and combined with the difference in LBP values ​​of each region in all adjacent frames of the current video, the multi-frame target factor of each region is obtained. Based on the target presence coefficient of each region in each frame image and the target factor of multiple frames, the correction coefficient of each region in each frame image is obtained. Then, it is determined whether each region in each frame image needs to be segmented into smaller regions. When segmentation is required, the side length of the segmented smaller regions in each frame image is obtained, and then the color-corrected video is obtained.

2. The color correction method for an ultra-high definition video camera according to claim 1, wherein, The specific process of clustering all pixels in each region is as follows: the average value of the R, G, and B channels of each pixel is recorded as the color value of each pixel; the color values ​​of all pixels in each region are used as the input of the DPC density clustering algorithm, and the absolute difference between the color values ​​of pixels is used as the distance metric, and multiple clusters are output.

3. The color correction method for an ultra-high definition video camera according to claim 1, wherein, The method for obtaining the color block difference degree of each region in each frame image is as follows: the minimum difference of the mean color values ​​of pixels in all two adjacent clusters in each region of each frame image is calculated and used as the color block difference degree of each region in each frame image.

4. The color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The calculation formula of the target existence coefficient of each region in each frame image is: ; in the formula, is the target existence coefficient of the i-th region in the k-th frame image, is the edge density in the i-th region in the k-th frame image, wherein the ratio of the total number of edge pixel points in each region to the total number of all pixel points is taken as the edge density of each region, is the variance of the gray value mean of all edges of the i-th region in the k-th frame image, is the color block difference degree of the i-th region in the k-th frame image, is a preset constant.

5. The color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The formula for calculating the target region weight of each region is as follows: In the formula, Let i be the target region weight for the i-th region. This represents the total number of edges of all neighboring regions of the i-th region in the u-th frame of the current video. , are the average color values ​​of the i-th region in the u-th and (u-1)-th frames of the current video, respectively, where U is the total number of frames in the current video.

6. The color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The method for obtaining the multi-frame target factors of each region is as follows: the cumulative result of the difference of LBP values ​​in the same region in all adjacent frames of the current video is calculated. The multi-frame target factors of the same region are positively correlated with the target region weight of the same region and negatively correlated with the cumulative result.

7. The color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The correction coefficients for each region in each frame of the image are the normalized result of the product of the target presence coefficient of each region and the target factor of multiple frames.

8. The color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The specific process for determining whether each region of each frame image needs to be divided into smaller regions is as follows: when the correction coefficient of any region of each frame image is greater than or equal to the preset segmentation threshold, it is determined that the region needs to be divided into smaller regions; otherwise, it is determined that the region does not need to be divided into smaller regions.

9. A color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The formula for calculating the side length of the small region after segmentation in each frame image is as follows: In the formula, Let v be the side length of the sub-region after segmenting the i-th region in the k-th frame image, and v be the preset initial side length. Let be the correction coefficient for the i-th region in the k-th frame image, and e be a preset coefficient. , This is the floor function.

10. A color correction method for an ultra-high-definition camera as described in claim 1, characterized in that, The specific process of obtaining the color-corrected video is as follows: each region is divided into small regions according to the side length of the corresponding small regions; each frame of the video data to be color-corrected uses the undivided regions and the divided small regions as inputs, and uses the histogram equalization algorithm to perform color correction on each region or small region, outputting each region or small region after color correction, and merging all the color-corrected regions and small regions in each frame to obtain a video in which each frame has been color-corrected.

Images