Video monitoring image optimization processing method based on low-illumination image enhancement algorithm

Abstract

The invention discloses a video monitoring image optimization processing method based on a low-illumination image enhancement algorithm, and relates to the technical field of image processing, and the method comprises the steps: S1, carrying out the screening processing of a video monitoring image, and obtaining a low-illumination video monitoring image; s2, acquiring an image enhancement result of the low-illumination video monitoring image based on a low-illumination image enhancement algorithm according to the low-illumination video monitoring image; and S3, performing optimization processing according to the image enhancement result of the low-illumination video monitoring image to obtain a video monitoring image optimization processing result. According to the method, a dark channel prior algorithm and various image enhancement technologies are combined, so that the brightness, contrast and definition of the low-illumination video monitoring image can be effectively improved, detail information in the image is richer, the definition and identifiability of the image are remarkably improved, and the image quality is improved. And more accurate and reliable image information is provided for safety monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a video surveillance image optimization processing method based on a low-light image enhancement algorithm. Background Art

[0002] With the widespread application of video surveillance technology, the quality of surveillance images under low-light conditions has received increasing attention. Traditional video surveillance systems often obtain low-light images due to insufficient light in low-light environments or in low-light environments, such as at night, on cloudy days, or in dimly lit indoors. These low-light images have problems such as low contrast, blurred details, and obvious noise, which seriously affect the quality and usability of surveillance images and bring great inconvenience to security monitoring work. Therefore, there is an urgent need for a video surveillance image optimization processing method based on a low-light image enhancement algorithm to solve the problem of poor image quality obtained by existing video surveillance systems in low-light environments. Summary of the invention

[0003] The purpose of the present invention is to propose a video surveillance image optimization processing method based on a low-light image enhancement algorithm, to optimize the video surveillance image through an effective low-light image enhancement method, to improve the clarity and recognizability of the image, thereby improving the performance and reliability of the video surveillance system.

[0004] To achieve the above object, the present invention provides a video surveillance image optimization processing method based on a low-light image enhancement algorithm, comprising the following steps: S1. Filtering and processing the video surveillance image to obtain a low-light video surveillance image; S2. Obtaining an image enhancement result of the low-light video surveillance image based on a low-light image enhancement algorithm according to the low-light video surveillance image; S3. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain a video surveillance image optimization processing result.

[0005] Optionally, the step of using the video surveillance image to perform screening processing to obtain the low-light video surveillance image includes: S1-1, using a video surveillance device to collect video surveillance images in real time; S1-2, acquiring historical low-light video surveillance images according to the video surveillance images and corresponding historical video surveillance images to construct a historical low-light video surveillance image dataset; S1-3, based on the video surveillance image, using the historical low-light video surveillance image data set to perform screening processing to obtain the low-light video surveillance image; The video surveillance images include low-light, normal-light and overexposed video surveillance images.

[0006] Optionally, performing screening processing based on the video surveillance image and utilizing the historical low-light video surveillance image dataset to obtain the low-light video surveillance image includes: S1-3-1. Using the historical low-light video surveillance image data set to obtain the brightness standard, contrast standard and color feature standard of the historical low-light video surveillance images as screening criteria for the low-light video surveillance images; S1-3-2, using the screening criteria for the low-light video surveillance images to screen the video surveillance images to obtain an initial low-light video surveillance image; S1-3-3, acquiring the low-light video surveillance image according to the screening criteria of the low-light video surveillance image and the initial low-light video surveillance image; Among them, the brightness standard is that the grayscale value of the low-light video surveillance image is less than 50, the contrast standard is that the contrast of the low-light video surveillance image is lower than 15, and the color feature standard is that the hue value of the low-light video surveillance image is 200-300 degrees.

[0007] Optionally, acquiring the low-light video surveillance image according to the screening criteria of the low-light video surveillance image and the initial low-light video surveillance image includes: S1-3-3-1. Perform grayscale processing on the initial low-light video surveillance image to obtain a grayscale value of the initial low-light video surveillance image as a brightness feature of the initial low-light video surveillance image; S1-3-3-2. According to the initial low-light video surveillance image, a Soble operator is used to obtain a gradient amplitude of the initial low-light video surveillance image as a contrast feature of the initial low-light video surveillance image; S1-3-3-3. Convert the initial low-light video surveillance image from the RGB color space to the HSV color space to obtain a hue value of the initial low-light video surveillance image as a color feature of the initial low-light video surveillance image; S1-3-3-4, determine whether the brightness feature of the initial low-light video surveillance image meets the brightness standard, if so, execute S1-3-3-5, otherwise, use the brightness feature that does not meet the standard to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-5, determine whether the contrast feature of the initial low-light video surveillance image meets the contrast standard, if so, execute S1-3-3-6, otherwise, use the non-compliant contrast feature to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-6. Determine whether the color features of the initial low-light video surveillance image meet the color feature standards. If so, obtain the initial low-light video surveillance image as the low-light video surveillance image. Otherwise, use the non-compliant color features to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2.

[0008] Optionally, obtaining an image enhancement result of the low-light video surveillance image based on a low-light image enhancement algorithm according to the low-light video surveillance image includes: S2-1, obtaining an illumination component image of a low-light video surveillance grayscale image based on the low-light video surveillance image based on a dark channel priori algorithm; S2-2, using the illumination component image of the low-light video surveillance grayscale image to obtain a reflection component image of the low-light video surveillance grayscale image; S2-3, performing weighted fusion on the illumination component image of the low-light video surveillance grayscale image and the reflection component image of the low-light video surveillance grayscale image to obtain a fusion result of the low-light video surveillance grayscale image as an enhanced image of the low-light video surveillance grayscale image; S2-4. Perform color restoration processing on the enhanced image of the low-light video surveillance grayscale image to obtain an image enhancement result of the low-light video surveillance image.

[0009] Optionally, obtaining an illumination component image of a low-light video surveillance grayscale image by using a dark channel priori algorithm based on the low-light video surveillance image includes: S2-1-1, using the low-light video surveillance image to obtain a low-light video surveillance grayscale image using a weighted average method; S2-1-2, performing division processing on the low-light video surveillance grayscale image to obtain a plurality of local areas; S2-1-3. Obtaining the minimum pixel value of each local area according to the plurality of local areas as the pixel value of the dark channel video surveillance image; S2-1-4, obtaining an atmospheric light value using the low-light video surveillance grayscale image based on the pixel value of the dark channel video surveillance image; S2-1-5. Obtain a preliminary prediction value of the illumination component according to the weight coefficient, the pixel value of the low-light video surveillance grayscale image and the atmospheric light value; S2-1-6. Performing Gaussian filtering according to the preliminary predicted value of the illumination component to obtain an illumination prediction result of the low-light video surveillance grayscale image as an illumination component image of the low-light video surveillance grayscale image; The weight coefficient is a difference adjustment parameter between the pixel value of the low-light video surveillance grayscale image and the atmospheric light value.

[0010] Optionally, obtaining a reflection component image of a low-light video surveillance grayscale image by using the illumination component image of the low-light video surveillance grayscale image includes: S2-2-1, using the illumination component image of the low-light video surveillance grayscale image and the low-light video surveillance grayscale image to obtain a reflection component prediction image of the low-light video surveillance grayscale image; S2-2-2, performing contrast stretching according to the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image; S2-2-3. According to the initial reflection component image of the low-light video surveillance grayscale image, using the Laplace operator to obtain the Laplace response value of the corresponding pixel of the initial reflection component image as the grayscale change value of the pixel of the initial reflection component image; S2-2-4, setting a grayscale change threshold of the initial reflection component image pixel according to the grayscale change value of the initial reflection component image pixel; S2-2-5. Perform edge detection on the low-light video surveillance grayscale image according to the grayscale change value of the initial reflection component image pixel and the grayscale change threshold of the initial reflection component image pixel to obtain the edge area and non-edge area of ​​the low-light video surveillance grayscale image as the edge detection result; S2-2-6. Perform detail enhancement processing on the initial reflection component image of the low-light video surveillance grayscale image according to the edge detection result to obtain a detail enhancement processing result of the initial reflection component image as the reflection component image of the low-light video surveillance grayscale image; Among them, the edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is greater than the grayscale change threshold of the initial reflection component image pixels, and the non-edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is less than or equal to the grayscale change threshold of the initial reflection component image pixels, and the grayscale change threshold of the initial reflection component image pixels is 20.

[0011] Optionally, performing contrast stretching according to the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image includes: S2-2-2-1. Using the reflection component prediction image of the low-light video surveillance grayscale image to divide and obtain a plurality of local image blocks of the reflection component prediction image; S2-2-2-2. Obtaining the grayscale variance of each local image block according to the local image blocks of the reflection component prediction image; S2-2-2-3. Obtaining the average grayscale variance of the reflection component prediction image based on the grayscale variance of each local image block; S2-2-2-4, setting a grayscale variance threshold of the reflection component prediction image according to the grayscale variance average value of the reflection component prediction image; S2-2-2-5, determine whether the grayscale variance averages of the reflection component prediction image meet the grayscale variance threshold of the reflection component prediction image, if so, execute S2-2-2-6, otherwise, directly execute S2-2-2-7; S2-2-2-6. Perform linear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image. S2-2-2-7. Perform nonlinear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image.

[0012] Optionally, performing optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result includes: S3-1, using the historical low-light video surveillance image to obtain an image enhancement result corresponding to the historical low-light video surveillance image; S3-2, setting an image enhancement standard for the historical low-light video surveillance image according to the image enhancement result of the historical low-light video surveillance image; S3-3, determining whether the image enhancement result of the low-light video surveillance image meets the image enhancement standard of the historical low-light video surveillance image, and if so, obtaining the image enhancement result of the low-light video surveillance image as the video surveillance image optimization processing result, otherwise, executing S3-4; S3-4. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result.

[0013] Optionally, performing optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result includes: S3-4-1. Obtain a standard enhanced image of a historical low-light video surveillance image using an image enhancement standard for historical low-light video surveillance images; S3-4-2, taking the standard enhanced image of the historical low-light video surveillance image as input, and the optimization processing result corresponding to the standard enhanced image as output, and constructing a video surveillance image enhancement model based on deep learning; S3-4-3, performing optimization processing using the video surveillance image enhancement model according to the image enhancement result of the low-light video surveillance image to obtain an optimization processing result of the image enhancement result; S3-4-4. Determine whether the optimized processing result of the image enhancement result meets the image enhancement standard of the historical low-light video surveillance image. If so, obtain the optimized processing result of the image enhancement result as the optimized processing result of the video surveillance image. Otherwise, obtain the optimized processing result of the image enhancement result that does not meet the standard and add it to the video surveillance image, and return to S1-2.

[0014] Compared with the closest prior art, the present invention has the following beneficial effects: The present invention can effectively improve the brightness, contrast and clarity of low-light video surveillance images by combining a dark channel prior algorithm with a variety of image enhancement technologies, making the detail information in the image richer, thereby significantly improving the clarity and recognizability of the image, and providing more accurate and reliable image information for security monitoring; image fusion and color restoration processing can make the enhanced image more natural and realistic, consistent with the color style of the original image, improve the practicality and aesthetics of the image, and facilitate monitoring personnel to observe and analyze image information; the image enhancement results of the low-light video surveillance images are optimized using a video surveillance image enhancement model, and the optimization processing results that do not meet the image enhancement standards are iteratively processed, thereby improving the processing speed as well as the clarity and reliability of the image, and ensuring the image enhancement effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 The present invention is a flowchart of a method for optimizing and processing video surveillance images based on a low-light image enhancement algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in combination with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] The terms used in the embodiments of the present invention are only used to explain the specific embodiments of the present invention and are not intended to limit the present invention.

[0019] like Figure 1 As shown, the embodiment of the present invention provides a video surveillance image optimization processing method based on a low-light image enhancement algorithm, comprising the following steps: S1. Filtering and processing the video surveillance image to obtain a low-light video surveillance image; S2. Obtaining an image enhancement result of the low-light video surveillance image based on a low-light image enhancement algorithm according to the low-light video surveillance image; S3. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain a video surveillance image optimization processing result.

[0020] S1 specifically includes: S1-1, using a video surveillance device to collect video surveillance images in real time; S1-2, acquiring historical low-light video surveillance images according to the video surveillance images and corresponding historical video surveillance images to construct a historical low-light video surveillance image dataset; S1-3, based on the video surveillance image, using the historical low-light video surveillance image data set to perform screening processing to obtain the low-light video surveillance image; The video surveillance images include low-light, normal-light and overexposed video surveillance images.

[0021] S1-3 specifically includes: S1-3-1. Using the historical low-light video surveillance image data set to obtain the brightness standard, contrast standard and color feature standard of the historical low-light video surveillance images as the screening standard for the low-light video surveillance images, wherein the brightness standard in this embodiment is set as the grayscale value of the low-light video surveillance image is less than 50, the contrast standard is set as the contrast of the low-light video surveillance image is less than 15, and the color feature standard is set as the hue value of the low-light video surveillance image is 200-300 degrees; S1-3-2, using the screening criteria for the low-light video surveillance images to screen the video surveillance images to obtain an initial low-light video surveillance image; S1-3-3. Acquire the low-light video surveillance image according to the screening criteria of the low-light video surveillance image and the initial low-light video surveillance image.

[0022] S1-3-3 specifically includes: S1-3-3-1. Perform grayscale processing on the initial low-light video surveillance image to obtain a grayscale value of the initial low-light video surveillance image as a brightness feature of the initial low-light video surveillance image; S1-3-3-2. According to the initial low-light video surveillance image, a Soble operator is used to obtain a gradient amplitude of the initial low-light video surveillance image as a contrast feature of the initial low-light video surveillance image; S1-3-3-3. Convert the initial low-light video surveillance image from the RGB color space to the HSV color space to obtain a hue value of the initial low-light video surveillance image as a color feature of the initial low-light video surveillance image; S1-3-3-4, determine whether the brightness feature of the initial low-light video surveillance image meets the brightness standard, if so, execute S1-3-3-5, otherwise, use the brightness feature that does not meet the standard to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-5, determine whether the contrast feature of the initial low-light video surveillance image meets the contrast standard, if so, execute S1-3-3-6, otherwise, use the non-compliant contrast feature to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-6. Determine whether the color features of the initial low-light video surveillance image meet the color feature standards. If so, obtain the initial low-light video surveillance image as the low-light video surveillance image. Otherwise, use the non-compliant color features to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2.

[0023] S2 specifically includes: S2-1, obtaining an illumination component image of a low-light video surveillance grayscale image based on the low-light video surveillance image based on a dark channel priori algorithm; S2-2, using the illumination component image of the low-light video surveillance grayscale image to obtain a reflection component image of the low-light video surveillance grayscale image; S2-3, performing weighted fusion on the illumination component image of the low-light video surveillance grayscale image and the reflection component image of the low-light video surveillance grayscale image to obtain a fusion result of the low-light video surveillance grayscale image as an enhanced image of the low-light video surveillance grayscale image; S2-4. Perform color restoration processing on the enhanced image of the low-light video surveillance grayscale image to obtain an image enhancement result of the low-light video surveillance image.

[0024] S2-1 specifically includes: S2-1-1, using the low-light video surveillance image to obtain a low-light video surveillance grayscale image using a weighted average method; S2-1-2, performing division processing on the low-light video surveillance grayscale image to obtain a plurality of local areas, in this embodiment, a square area of ​​15×15 pixels is used as a sub-block (the size can be adjusted according to the actual image resolution and effect); S2-1-3. Obtaining the minimum pixel value of each local area according to the plurality of local areas as the pixel value of the dark channel video surveillance image; S2-1-4, based on the pixel value of the dark channel video surveillance image, using the low-light video surveillance grayscale image to obtain the atmospheric light value, specifically: Setting a target pixel point of a dark channel video surveillance image, wherein the atmospheric light value is obtained at the target pixel point of the dark channel video surveillance image. In this embodiment, the atmospheric light value is set to be obtained at a brighter 0.1% (the ratio can be fine-tuned) pixel point in the dark channel video surveillance image; Obtaining a pixel value corresponding to the target pixel in the low-light video surveillance grayscale image according to the target pixel of the dark channel video surveillance image and the pixel value of the dark channel video surveillance image; Obtaining a maximum pixel value as an atmospheric light value according to a corresponding pixel value of the target pixel point in the low-light video surveillance grayscale image; S2-1-5. Obtain a preliminary prediction value of the illumination component according to the weight coefficient, the pixel value of the low-light video surveillance grayscale image and the atmospheric light value, specifically: First, a difference adjustment parameter between the pixel value of the low-light video surveillance grayscale image and the atmospheric light value is set as a weight coefficient according to the pixel value of the low-light video surveillance grayscale image and the atmospheric light value, wherein when the weight coefficient is closer to 1, it means that the difference adjustment between the pixel value of the low-light video surveillance grayscale image and the atmospheric light value is smaller, and it is more inclined to retain the relative proportion of the original illumination information, so that when processing some low-light images with relatively uniform illumination and rich details in the subsequent processing, the original light and shadow structure of the image can be better restored to avoid image distortion caused by over-correction; Then, the preliminary prediction value of the illumination component is calculated according to the weight coefficient in combination with the pixel value of the low-light video surveillance grayscale image and the atmospheric light value. The preliminary prediction value J(x) of the illumination component is:

[0025] Where A is the atmospheric light value, is a weight coefficient, which is generally set to a value of 0.95 to 1. In this embodiment, the weight coefficient is set to 0.95. I(x) is the pixel value of the grayscale image of the low-light video surveillance; S2-1-6. Perform Gaussian filtering according to the preliminary predicted value of the illumination component to obtain the illumination prediction result of the low-light video surveillance grayscale image as the illumination component image of the low-light video surveillance grayscale image. In this embodiment, the standard deviation of the Gaussian filter is selected between 10-30.

[0026] S2-2 specifically includes: S2-2-1, using the illumination component image of the low-light video surveillance grayscale image and the low-light video surveillance grayscale image to obtain a reflection component prediction image of the low-light video surveillance grayscale image; S2-2-2, performing contrast stretching according to the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image; S2-2-3. According to the initial reflection component image of the low-light video surveillance grayscale image, using the Laplace operator to obtain the Laplace response value of the corresponding pixel of the initial reflection component image as the grayscale change value of the pixel of the initial reflection component image; S2-2-4, setting a grayscale change threshold of the initial reflection component image pixel according to the grayscale change value of the initial reflection component image pixel; S2-2-5. Perform edge detection on the low-light video surveillance grayscale image according to the grayscale change value of the initial reflection component image pixel and the grayscale change threshold of the initial reflection component image pixel to obtain the edge area and non-edge area of ​​the low-light video surveillance grayscale image as the edge detection result; S2-2-6. Perform detail enhancement processing on the initial reflection component image of the low-light video surveillance grayscale image according to the edge detection result to obtain a detail enhancement processing result of the initial reflection component image as the reflection component image of the low-light video surveillance grayscale image; Among them, the edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is greater than the grayscale change threshold of the initial reflection component image pixels, and the non-edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is less than or equal to the grayscale change threshold of the initial reflection component image pixels.

[0027] In the monitoring image of clear building outlines under low light, due to the low noise and obvious edge features, a higher threshold is set, that is, the grayscale change threshold is set to 50, which can effectively eliminate interference and accurately identify the real building edge; while in the monitoring image of low light and foggy roads, there is a lot of noise and blurred edges. Lowering the threshold, that is, setting the grayscale change threshold to 20, can capture more real roads, vehicles, etc. covered by fog. In this embodiment, the grayscale change threshold of the initial reflection component image pixel is set to 20.

[0028] A high-frequency boost filtering algorithm is used in the edge area of ​​the low-light video surveillance grayscale image, and a local adaptive histogram equalization algorithm is used in the non-edge area of ​​the low-light video surveillance grayscale image. The high-frequency boost filtering and local adaptive histogram equalization algorithms can perform targeted enhancement processing on different areas of the image, making the edges of the image clearer and sharper, and further improving the local contrast, thereby enhancing the overall visual effect of the image.

[0029] S2-2-2 specifically includes: S2-2-2-1. Using the reflection component prediction image of the low-light video surveillance grayscale image to divide and obtain a plurality of local image blocks of the reflection component prediction image; S2-2-2-2. Obtaining the grayscale variance of each local image block according to the local image blocks of the reflection component prediction image; S2-2-2-3. Obtain the average grayscale variance of the reflection component prediction image based on the grayscale variance of each local image block, specifically: The grayscale variance of all local image blocks is added and divided by the total number of image blocks to calculate the average grayscale variance of the reflection component prediction image. ; S2-2-2-4. Setting a grayscale variance threshold of the reflection component prediction image according to the grayscale variance average value of the reflection component prediction image. Specifically: According to the grayscale variance average of the reflection component prediction image, the grayscale variance threshold of the reflection component prediction image is set to , k is a constant greater than 1, generally between 1.5 and 3; S2-2-2-5, determine whether the grayscale variance averages of the reflection component prediction image meet the grayscale variance threshold of the reflection component prediction image, if so, execute S2-2-2-6, otherwise, directly execute S2-2-2-7; S2-2-2-6, performing linear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image, specifically: Obtaining all pixels of the reflection component prediction image according to the reflection component prediction image of the low-light video surveillance grayscale image, and calculating a minimum grayscale value and a maximum grayscale value, which define a grayscale range of the reflection component prediction image; The target range to which the grayscale range of the image is stretched is determined according to actual needs, which is usually the full range from 0 to 255, that is, the minimum target grayscale value is 0 and the maximum target grayscale value is 255. In this embodiment, the target grayscale range is set according to the grayscale range of the image predicted by the reflection component; Performing linear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image based on the linear transformation coefficient to obtain an initial reflection component image of the low-light video surveillance grayscale image; S2-2-2-7, performing nonlinear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image, specifically: Acquire all pixels of the reflection component prediction image according to the reflection component prediction image of the low-light video surveillance grayscale image, and normalize the grayscale values ​​to the interval [0, 1] to acquire the normalized grayscale values ​​of the reflection component prediction image; A nonlinear transformation formula is used based on the normalized grayscale value of the reflection component prediction image and the nonlinear transformation coefficient to obtain the nonlinear transformation grayscale value of the reflection component prediction image, wherein the nonlinear transformation coefficient is selected or adjusted according to the actual situation of the reflection component prediction image; The nonlinear transformation grayscale value of the reflection component prediction image is denormalized to the interval [0, 255] to obtain the initial reflection component image of the low-light video surveillance grayscale image, completing the nonlinear contrast stretching.

[0030] S3 specifically includes: S3-1, using the historical low-light video surveillance image to obtain an image enhancement result corresponding to the historical low-light video surveillance image; S3-2, setting an image enhancement standard for the historical low-light video surveillance image according to the image enhancement result of the historical low-light video surveillance image; S3-3, determining whether the image enhancement result of the low-light video surveillance image meets the image enhancement standard of the historical low-light video surveillance image, and if so, obtaining the image enhancement result of the low-light video surveillance image as the video surveillance image optimization processing result, otherwise, executing S3-4; S3-4. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result.

[0031] S3-4 specifically includes: S3-4-1. Obtain a standard enhanced image of a historical low-light video surveillance image using an image enhancement standard for historical low-light video surveillance images; S3-4-2, taking the standard enhanced image of the historical low-light video surveillance image as input, and the optimization processing result corresponding to the standard enhanced image as output, and constructing a video surveillance image enhancement model based on deep learning; S3-4-3, performing optimization processing using the video surveillance image enhancement model according to the image enhancement result of the low-light video surveillance image to obtain an optimization processing result of the image enhancement result; S3-4-4. Determine whether the optimized processing result of the image enhancement result meets the image enhancement standard of the historical low-light video surveillance image. If so, obtain the optimized processing result of the image enhancement result as the optimized processing result of the video surveillance image. Otherwise, obtain the optimized processing result of the image enhancement result that does not meet the standard and add it to the video surveillance image, and return to S1-2.

[0032] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0033] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0034] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0035] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A video surveillance image optimization processing method based on a low-light image enhancement algorithm, characterized in that: The specific steps include: S1. Filtering and processing the video surveillance image to obtain a low-light video surveillance image; S2. Obtaining an image enhancement result of the low-light video surveillance image based on a low-light image enhancement algorithm according to the low-light video surveillance image; S3. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain a video surveillance image optimization processing result.

2. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 1 is characterized in that: The low-light video surveillance images are obtained by screening and processing the video surveillance images, including: S1-1, using a video surveillance device to collect video surveillance images in real time; S1-2, acquiring historical low-light video surveillance images according to the video surveillance images and corresponding historical video surveillance images to construct a historical low-light video surveillance image dataset; S1-3, based on the video surveillance image, using the historical low-light video surveillance image data set to perform screening processing to obtain the low-light video surveillance image; The video surveillance images include low-light, normal-light and overexposed video surveillance images.

3. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 2 is characterized in that: The method of obtaining the low-light video surveillance image by filtering and processing the historical low-light video surveillance image dataset based on the video surveillance image includes: S1-3-1. Using the historical low-light video surveillance image data set to obtain the brightness standard, contrast standard and color feature standard of the historical low-light video surveillance images as screening criteria for the low-light video surveillance images; S1-3-2, using the screening criteria for the low-light video surveillance images to screen the video surveillance images to obtain an initial low-light video surveillance image; S1-3-3, acquiring the low-light video surveillance image according to the screening criteria of the low-light video surveillance image and the initial low-light video surveillance image; Among them, the brightness standard is that the grayscale value of the low-light video surveillance image is less than 50, the contrast standard is that the contrast of the low-light video surveillance image is lower than 15, and the color feature standard is that the hue value of the low-light video surveillance image is 200-300 degrees.

4. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 3 is characterized in that: Acquiring the low-light video surveillance image according to the screening criteria of the low-light video surveillance image and the initial low-light video surveillance image includes: S1-3-3-1. Perform grayscale processing on the initial low-light video surveillance image to obtain a grayscale value of the initial low-light video surveillance image as a brightness feature of the initial low-light video surveillance image; S1-3-3-2. Using the Soble operator according to the initial low-light video surveillance image, obtain the gradient amplitude of the initial low-light video surveillance image as a contrast feature of the initial low-light video surveillance image; S1-3-3-3. Convert the initial low-light video surveillance image from the RGB color space to the HSV color space to obtain a hue value of the initial low-light video surveillance image as a color feature of the initial low-light video surveillance image; S1-3-3-4, determine whether the brightness feature of the initial low-light video surveillance image meets the brightness standard, if so, execute S1-3-3-5, otherwise, use the brightness feature that does not meet the standard to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-5, determine whether the contrast feature of the initial low-light video surveillance image meets the contrast standard, if so, execute S1-3-3-6, otherwise, use the non-compliant contrast feature to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2; S1-3-3-6. Determine whether the color features of the initial low-light video surveillance image meet the color feature standards. If so, obtain the initial low-light video surveillance image as the low-light video surveillance image. Otherwise, use the non-compliant color features to obtain the corresponding initial low-light video surveillance image, add it to the video surveillance image, and return to S1-2.

5. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 1 is characterized in that: Obtaining an image enhancement result of the low-light video surveillance image based on the low-light image enhancement algorithm according to the low-light video surveillance image includes: S2-1, obtaining an illumination component image of a low-light video surveillance grayscale image based on the low-light video surveillance image based on a dark channel priori algorithm; S2-2, using the illumination component image of the low-light video surveillance grayscale image to obtain a reflection component image of the low-light video surveillance grayscale image; S2-3, performing weighted fusion on the illumination component image of the low-light video surveillance grayscale image and the reflection component image of the low-light video surveillance grayscale image to obtain a fusion result of the low-light video surveillance grayscale image as an enhanced image of the low-light video surveillance grayscale image; S2-4. Perform color restoration processing on the enhanced image of the low-light video surveillance grayscale image to obtain an image enhancement result of the low-light video surveillance image.

6. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 5 is characterized in that: Acquiring an illumination component image of a low-light video surveillance grayscale image using a dark channel priori algorithm based on the low-light video surveillance image includes: S2-1-1, using the low-light video surveillance image to obtain a low-light video surveillance grayscale image using a weighted average method; S2-1-2, performing division processing on the low-light video surveillance grayscale image to obtain a plurality of local areas; S2-1-3. Obtaining the minimum pixel value of each local area according to the plurality of local areas as the pixel value of the dark channel video surveillance image; S2-1-4, obtaining an atmospheric light value using the low-light video surveillance grayscale image based on the pixel value of the dark channel video surveillance image; S2-1-5. Obtain a preliminary prediction value of the illumination component according to the weight coefficient, the pixel value of the low-light video surveillance grayscale image and the atmospheric light value; S2-1-6. Performing Gaussian filtering according to the preliminary predicted value of the illumination component to obtain an illumination prediction result of the low-light video surveillance grayscale image as an illumination component image of the low-light video surveillance grayscale image; The weight coefficient is a difference adjustment parameter between the pixel value of the low-light video surveillance grayscale image and the atmospheric light value.

7. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 6 is characterized in that: Using the illumination component image of the low-light video surveillance grayscale image to obtain the reflection component image of the low-light video surveillance grayscale image includes: S2-2-1, using the illumination component image of the low-light video surveillance grayscale image and the low-light video surveillance grayscale image to obtain a reflection component prediction image of the low-light video surveillance grayscale image; S2-2-2, performing contrast stretching according to the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image; S2-2-3. According to the initial reflection component image of the low-light video surveillance grayscale image, using the Laplace operator to obtain the Laplace response value of the corresponding pixel of the initial reflection component image as the grayscale change value of the pixel of the initial reflection component image; S2-2-4, setting a grayscale change threshold of the initial reflection component image pixel according to the grayscale change value of the initial reflection component image pixel; S2-2-5. Perform edge detection on the low-light video surveillance grayscale image according to the grayscale change value of the initial reflection component image pixel and the grayscale change threshold of the initial reflection component image pixel to obtain the edge area and non-edge area of ​​the low-light video surveillance grayscale image as the edge detection result; S2-2-6. Perform detail enhancement processing on the initial reflection component image of the low-light video surveillance grayscale image according to the edge detection result to obtain a detail enhancement processing result of the initial reflection component image as the reflection component image of the low-light video surveillance grayscale image; Among them, the edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is greater than the grayscale change threshold of the initial reflection component image pixels, and the non-edge area of ​​the low-light video surveillance grayscale image is the pixels whose grayscale change value of the initial reflection component image pixels is less than or equal to the grayscale change threshold of the initial reflection component image pixels, and the grayscale change threshold of the initial reflection component image pixels is 20.

8. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 7 is characterized in that: Performing contrast stretching according to the reflection component prediction image of the low-light video surveillance grayscale image to obtain the initial reflection component image of the low-light video surveillance grayscale image includes: S2-2-2-1. Using the reflection component prediction image of the low-light video surveillance grayscale image to divide and obtain a plurality of local image blocks of the reflection component prediction image; S2-2-2-2. Obtaining the grayscale variance of each local image block according to the local image blocks of the reflection component prediction image; S2-2-2-3. Obtaining the average grayscale variance of the reflection component prediction image based on the grayscale variance of each local image block; S2-2-2-4, setting a grayscale variance threshold of the reflection component prediction image according to the grayscale variance average value of the reflection component prediction image; S2-2-2-5, determine whether the grayscale variance averages of the reflection component prediction image meet the grayscale variance threshold of the reflection component prediction image, if so, execute S2-2-2-6, otherwise, directly execute S2-2-2-7; S2-2-2-6. Perform linear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image. S2-2-2-7. Perform nonlinear contrast stretching on the reflection component prediction image of the low-light video surveillance grayscale image to obtain an initial reflection component image of the low-light video surveillance grayscale image.

9. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 2 is characterized in that: Performing optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result includes: S3-1, using the historical low-light video surveillance image to obtain an image enhancement result corresponding to the historical low-light video surveillance image; S3-2, setting an image enhancement standard for the historical low-light video surveillance image according to the image enhancement result of the historical low-light video surveillance image; S3-3, determining whether the image enhancement result of the low-light video surveillance image meets the image enhancement standard of the historical low-light video surveillance image, and if so, obtaining the image enhancement result of the low-light video surveillance image as the video surveillance image optimization processing result, otherwise, executing S3-4; S3-4. Perform optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result.

10. The video surveillance image optimization processing method based on the low-light image enhancement algorithm according to claim 9 is characterized in that: Performing optimization processing according to the image enhancement result of the low-light video surveillance image to obtain the video surveillance image optimization processing result includes: S3-4-1. Obtain a standard enhanced image of a historical low-light video surveillance image using an image enhancement standard for historical low-light video surveillance images; S3-4-2, taking the standard enhanced image of the historical low-light video surveillance image as input, and the optimization processing result corresponding to the standard enhanced image as output, and constructing a video surveillance image enhancement model based on deep learning; S3-4-3, performing optimization processing using the video surveillance image enhancement model according to the image enhancement result of the low-light video surveillance image to obtain an optimization processing result of the image enhancement result; S3-4-4. Determine whether the optimized processing result of the image enhancement result meets the image enhancement standard of the historical low-light video surveillance image. If so, obtain the optimized processing result of the image enhancement result as the optimized processing result of the video surveillance image. Otherwise, obtain the optimized processing result of the image enhancement result that does not meet the standard and add it to the video surveillance image, and return to S1-2.

Images