Low illumination image and video enhancement method and device based on atmospheric scattering model

Abstract

The present invention discloses a method and device for enhancing low-illuminance images and videos based on an atmospheric scattering model. The method comprises: refining the V component based on a refinement equation, obtaining the transmittance to obtain the local maximum value of the V component, obtaining a local maximum map and refining it, obtaining the transmittance to obtain the final transmittance based on the transmittance and the superposition coefficient, selecting a preset proportion of pixels according to the brightness, obtaining the positions corresponding to the preset proportion of pixels, selecting the V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculating the average value of the V component value as the global atmospheric light; substituting the final transmittance and the global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, and refining it to obtain the final enhanced V component; converting the image from the HSV' color space to the RGB space to obtain an enhanced image, and synthesizing an enhanced video. The device comprises: a processor and a memory. The images and videos obtained by the present invention can effectively highlight the key information in the image.

Description

Technical Field

[0001] The present invention relates to low-illumination image and video enhancement technology, belongs to the field of image processing, and in particular to a low-illumination image and video enhancement method and device based on an atmospheric scattering model. Background Art

[0002] With the advancement of science and technology, surveillance cameras are widely used in various fields. However, in low-light conditions, the brightness and contrast of the video images captured by the cameras are low, and due to the influence of ambient light, the images are prone to uneven illumination such as extremely dark and extremely bright, which makes many key information in the video images submerged and cannot meet practical applications. In order to make better use of the video images obtained in low-light conditions, they need to be enhanced.

[0003] Researchers have proposed many algorithms for low-light video images. Early enhancement algorithms directly adjusted the pixel values ​​of the image to improve the brightness of the low-light image, such as histogram equalization algorithm and grayscale transformation algorithm, but the above algorithms have problems such as poor adaptability, limited enhancement ability, and easy color distortion.

[0004] In order to solve these defects, model-based algorithms have been proposed and widely used in image enhancement. The Retinex model is often used for color image enhancement. It can well preserve the color information of the image, but this type of algorithm is prone to over-enhancement for images with uneven illumination. The dehazing model is also a commonly used low-illumination image enhancement model. Based on the similarity between low-illumination images and foggy images, researchers have proposed many algorithms, but this type of algorithm is prone to introduce artifacts in the enhanced image due to model parameter estimation problems.

[0005] Although existing algorithms can effectively enhance low-light images, they have problems such as over-enhancement and introduction of halo artifacts, which causes loss of image details and cannot meet the needs of applications in complex low-light situations. Summary of the invention

[0006] The present invention provides a method and device for enhancing low-light images and videos based on an atmospheric scattering model. After processed by the present invention, the brightness of dark areas of low-light images and videos is effectively improved, while the brightness of bright areas is well maintained, and overexposure does not occur to lose details. The obtained images and videos are natural and have clear details, and the key information in the images can be effectively highlighted. See the following description for details:

[0007] In a first aspect, a low-illumination image and video enhancement method based on an atmospheric scattering model, the method comprising:

[0008] Refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance

[0009] Based on transmittance Transmittance And the superposition coefficient to obtain the final transmittance;

[0010] Selecting a preset proportion of pixels from the final transmittance according to the brightness, obtaining positions corresponding to the preset proportion of pixels, selecting V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculating the average value of the V component values ​​as the global atmospheric light;

[0011] Substitute the final transmittance and global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, which is then refined to obtain the final enhanced V component;

[0012] Keeping the hue and saturation components unchanged, the image is converted from the HSV' color space to the RGB space using MATLAB to obtain the enhanced image, and finally the enhanced video is synthesized.

[0013] In one embodiment, the V component is refined based on the refinement equation to obtain the transmittance Specifically:

[0014]

[0015] where α is the equilibrium parameter, ||·|| F and ||·|| 1 They refer to the Frobenious norm and l respectively. 1 norm, W is the weight matrix defined, Express First-order derivative filtering is performed in the horizontal and vertical directions.

[0016] In another embodiment, the weight matrix is:

[0017]

[0018]

[0019] Among them, G σ (x, y) is a Gaussian kernel function with standard deviation σ, v is the vertical direction, V(y) is the brightness of point y, W h (x) is the weight in the horizontal direction, W v (x) is the weight in the vertical direction; Ω(x) is the 3×3 neighborhood centered on x; h is the horizontal direction; ε is a constant close to 0 to prevent the denominator from being zero.

[0020] Wherein, the superposition coefficient is:

[0021]

[0022] Where max(V) is the maximum value of the V component, var(V) is the variance of the V component, and mean(V) is the mean of the V component.

[0023] Furthermore, the V component obtained by preliminary enhancement is refined to obtain the final enhanced V component as follows:

[0024] An S-shaped coefficient is defined using the clarity function to superimpose the high-frequency part representing detail information with the initially enhanced V component;

[0025] Among them, the S-type coefficient function is:

[0026]

[0027] Where k represents the slope of the control function, AG m represents the clarity function;

[0028] The final enhanced V component results are as follows:

[0029]

[0030] Among them, Hig(x,y) represents the high-frequency information of the image, Represents the V component of the initial enhancement.

[0031] The method of using MATLAB to convert the image from the HSV' color space to the RGB space to obtain an enhanced image and finally synthesize the enhanced video is as follows:

[0032] Using the hsv2rgb function in MATLAB, the image is converted from the HSV' color space to the RGB space to obtain an enhanced image while keeping the hue and saturation components unchanged;

[0033] Each subsequent frame is enhanced based on the average brightness of the first enhanced frame to ensure the consistency of the average brightness. The enhanced images are then read sequentially and synthesized into an enhanced video at a rate of 30 frames per second.

[0034] In a second aspect, a low-light image and video enhancement device based on an atmospheric scattering model, the device comprising:

[0035] The first acquisition module is used to refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance

[0036] The second acquisition module is used to obtain Transmittance And the superposition coefficient to obtain the final transmittance;

[0037] Generate a global atmospheric light module, which is used to select a preset proportion of pixels from the final transmittance according to the brightness, obtain the positions corresponding to the preset proportion of pixels, select V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculate the average value of the V component value as the global atmospheric light;

[0038] A refinement module is used to substitute the final transmittance and global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, and then refine it to obtain the final enhanced V component;

[0039] The synthesis module is used to keep the hue and saturation components unchanged, use MATLAB to convert the image from the HSV' color space to the RGB space to obtain the enhanced image, and finally synthesize the enhanced video.

[0040] In a third aspect, a low-light image and video enhancement device based on an atmospheric scattering model is provided, the device comprising: a processor and a memory, the memory storing program instructions, the processor calling the program instructions stored in the memory to enable the device to execute any one of the method steps described in the first aspect.

[0041] In a fourth aspect, a computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method steps in the first aspect.

[0042] The beneficial effects of the technical solution provided by the present invention are:

[0043] 1) The enhanced video image obtained by the present invention has natural brightness, high contrast and clarity, and is suitable for different low-light scenes, and is well applicable to overall dark images and images with uneven brightness;

[0044] 2) The present invention explains the atmospheric scattering model under low illumination conditions. The parameter estimation method proposed in the present invention can accurately estimate the atmospheric light value and transmittance, and obtain a more accurate illumination distribution, thereby effectively improving the brightness of the dark area of ​​the original video image while suppressing excessive enhancement of the bright area, thereby avoiding the loss of video image information;

[0045] 3) The detail optimization algorithm proposed by the present invention can significantly improve the edge and detail information of low-light video images. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The present invention is a flowchart of a low-light image and video enhancement method based on an atmospheric scattering model;

[0047] Figure 2 This is a schematic diagram of the experimental results of a single-frame image with low illumination that is darker overall;

[0048] Among them, (a) is the original image and (b) is the enhanced image.

[0049] Figure 3 This is a schematic diagram of the experimental results of a single-frame image with low illumination and uneven brightness;

[0050] Among them, (a) is the original image and (b) is the enhanced image.

[0051] Figure 4 It is a structural schematic diagram of a low-light image and video enhancement device based on an atmospheric scattering model;

[0052] Figure 5 Another structural schematic diagram of a low-illumination image and video enhancement device based on an atmospheric scattering model. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are described in further detail below.

[0054] Example 1

[0055] The experimental objects used in the embodiments of the present invention are low-light images and videos in different scenes. To solve the problems of uneven illumination and poor visibility in low-light videos, a low-light image and video enhancement method based on an atmospheric scattering model is proposed. Figure 1 , the method comprises the following steps:

[0056] 101: Refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance

[0057] 102: Based on transmittance Transmittance And the superposition coefficient to obtain the final transmittance;

[0058] 103: selecting a preset proportion of pixels from the final transmittance according to the brightness, obtaining positions corresponding to the preset proportion of pixels, selecting V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculating an average value of the V component values ​​as the global atmospheric light;

[0059] 104: Substituting the final transmittance and the global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, and performing refinement to obtain a final enhanced V component;

[0060] 105: Keeping the hue and saturation components unchanged, use MATLAB to convert the image from HSV' color space to RGB space to obtain the enhanced image, and finally synthesize the enhanced video.

[0061] The V component initially enhanced in step 104 is refined to obtain the final enhanced V component as follows:

[0062] An S-shaped coefficient is defined using the clarity function to superimpose the high-frequency part representing detail information with the initially enhanced V component;

[0063] Among them, the S-type coefficient function is:

[0064]

[0065] Where k represents the slope of the control function, AG m represents the clarity function;

[0066] The final enhanced V component results are as follows:

[0067]

[0068] Among them, Hig(x,y) represents the high-frequency information of the image, Represents the V component of the initial enhancement.

[0069] In summary, the images and videos obtained through the processing of steps 101 to 105 in the embodiment of the present invention are natural and have clear details, and can effectively highlight the key information in the image.

[0070] Example 2

[0071] Combine the following Figure 1 , calculation formula, specific examples are used to further illustrate the scheme in Example 1, as described below:

[0072] Step 201: using a MATLAB function to decompose the low-light video into multiple single-frame low-light images, and writing them into a fixed folder in sequence;

[0073] The above order is: the video is composed of frames of images in time sequence, so the order here is in time sequence.

[0074] Step 202: read the image, convert the image from RGB (red, green, blue) color space to HSV (hexagonal pyramid model) space using rgb2hsv function in MATLAB, and extract V component;

[0075] Among them, H is the hue component, S is the saturation component, and V is the lightness component.

[0076] Step 203: estimating transmittance;

[0077] According to the meaning of transmittance t(x), this parameter is the base layer of the low-light image and represents the illumination distribution of the image. Therefore, the transmittance estimation result should be as smooth as possible while retaining the overall structure of the low-light image.

[0078] Among them, the V component describes the brightness distribution of the image, so the V component can be refined using the following refinement equation to obtain the transmittance

[0079]

[0080] where α is the equilibrium parameter, ||·|| F and ||·|| 1 They refer to the Frobenious norm and l respectively. 1 norm, W is the weight matrix defined, Express At the level and vertical Direction first-order derivative filtering is performed. The first term of the above equation takes into account and V, the second term takes into account the fidelity between Structure-aware smoothness.

[0081] In the embodiment of the present invention, in order to accurately estimate the transmittance and thus reduce overexposure, the embodiment of the present invention redefines the weight matrix:

[0082]

[0083] Among them, G σ (x,y) is generated by a Gaussian kernel function with standard deviation σ, defined as follows:

[0084]

[0085] Where dist(x,y) measures the spatial Euclidean distance between x and y, ε is a small parameter to prevent the denominator from being zero, h is the horizontal direction, v is the vertical direction, V(y) is the brightness of point y, and W h (x) is the weight in the horizontal direction, W v (x) is the weight in the vertical direction. The weight here only needs to be calculated once without iteration, which reduces the time complexity.

[0086] The transmittance obtained by the above process is not accurate in estimating the bright area of ​​the image, and overexposure will still occur in the extremely bright area. Therefore, the local maximum value of the V component is calculated to obtain the local maximum map M(x):

[0087]

[0088] Wherein, Ω is the local block size, and the commonly used 3×3 is used in the embodiment of the present invention. Using the above refinement equation, the local maximum map M(x) is refined to obtain the transmittance

[0089] To avoid transmittance and The estimated transmittance of the stacked estimate overestimates the bright areas, and the following stacking coefficients are defined:

[0090]

[0091] Where max(V) is the maximum value of the V component, var(V) is the variance of the V component, and mean(V) is the mean of the V component.

[0092] The final transmittance t(x) is expressed as follows:

[0093]

[0094] in, is the refinement result of the V component, is the refinement result of the maximum map M(x).

[0095] Step 204: estimating global atmospheric light A;

[0096] Among them, the top 1% of pixels with the largest brightness are taken from the transmittance image t(x) estimated in step 203, the positions of these pixels are found, the values ​​of the V components of these pixels are taken from the low-illuminance image according to these positions, and the average value of all values ​​is calculated as the global atmospheric light A.

[0097] Step 205: Substitute the transmittance t(x) estimated in step 203 and the global atmospheric light A estimated in step 204 into the low-illuminance atmospheric scattering model to obtain a preliminarily enhanced V component.

[0098]

[0099] Among them, t 0 is a very small constant to prevent the denominator from being zero, and its value is 0.01.

[0100] The definition and physical meaning of the low-illumination atmospheric scattering model are explained as follows:

[0101] I(x)=J(x)t(x)+A(1-t(x)) (8)

[0102] Among them, I(x) is the input low-light image, J(x) represents the enhanced image, A is the atmospheric light in a dark environment, and t(x) is the transmittance, which describes the light reaching the camera, including ambient light.

[0103] Step 206: Refine the initially enhanced V component to obtain a final enhanced V component.

[0104] Low-light images have the characteristic that the lower the illumination, the worse the detail clarity. Based on this characteristic, an S-shaped coefficient is defined using the clarity function to separate the high-frequency part representing detail information from the V component that is initially enhanced. Superposition is performed to improve detail clarity, and the coefficient function is defined as follows:

[0105]

[0106] Where k represents the slope of the control function, AG m represents the clarity function, which is expressed as follows:

[0107]

[0108] Among them, M×N represents the image size, express The horizontal gradient, express Vertical gradient.

[0109] Therefore, the final enhanced V component results are as follows:

[0110]

[0111] Among them, Hig(x,y) represents the high-frequency information of the image, Represents the V component of the initial enhancement.

[0112] Step 207: using the hsv2rgb function in MATLAB, keeping the hue (H) and saturation (S) components unchanged, converting the image from the HSV' color space to the RGB space to obtain an enhanced image;

[0113] Among them, the hsv2rgb function is a well-known function in the art, and the embodiment of the present invention will not be described in detail.

[0114] This step 207 only enhances the V component. The enhanced V component is represented by V', and the hue (H) and saturation (S) components remain unchanged, that is, the HSV color space is transformed into the HSV' color space, and the image is converted from the HSV' color space to the RGB space to obtain an enhanced image.

[0115] Step 208: Enhance each subsequent frame based on the average brightness of the enhanced first frame to ensure the consistency of the average brightness and avoid flickering and other undesirable phenomena in the synthesized video. Then read the enhanced images sequentially and synthesize them into an enhanced video at a rate of 30 frames per second.

[0116] In summary, the images and videos obtained through the processing of steps 201 to 208 in the embodiment of the present invention are natural and clear in details, and can effectively highlight the key information in the image.

[0117] Example 3

[0118] Combined with Table 1-Table 3, Figure 2 and Figure 3 The feasibility of the schemes in Examples 1 and 2 is verified as follows:

[0119] In low-light situations, the lighting conditions are complex, and the video images captured by the camera equipment are prone to uneven brightness distribution such as extremely dark and extremely bright areas. Therefore, experiments are conducted on extremely dark images and images with extremely bright areas with uneven illumination. Figure 2 (a) shows a low-light image captured in an extremely dark situation. Figure 2 (b) shows the corresponding enhanced image; Figure 3 (a) shows an image acquired in a low-light situation with lights. Figure 3 (b) shows the corresponding enhanced image; Figure 2 (b) and Figure 3 (b) It can be seen that the proposed method can well enhance the brightness of the dark area of ​​the image while maintaining the brightness of the bright area of ​​the image, effectively enhancing the details and clarity of the image. Combined with the subjective effect, the enhancement results were objectively evaluated, and 7 indicators were calculated, including: 4 non-reference image evaluation indicators such as clarity (AG), entropy (H), contrast (C) and edge strength (e), and 3 reference image evaluation indicators such as block-based contrast (PCQI), peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). The larger the value of the 7 indicators tested, the higher the quality of the enhanced image. In addition, the running time of the algorithm is also recorded. Table 1 and Table 2 correspond to Figure 2 and Figure 3 objective evaluation results.

[0120] Table 1

[0121] AG H C e PCQI SSIM PSNR Times This method 15.23 7.08 1747.78 143.36 1.17 0.11 14.39 0.56

[0122] Table 2

[0123] AG H C e PCQI SSIM PSNR Times This method 15.39 7.57 1251.05 149.82 1.29 0.19 14.94 0.58

[0124] In order to prove the effectiveness and universality of the algorithm, the enhancement results of 200 low-light images taken on site were qualitatively analyzed, and the method was compared with four advanced enhancement methods. The average values ​​of various indicators were calculated and obtained in Table 3.

[0125] Table 3

[0126] method AG H C e PCQI SSIM PSNR Times ALSM 8.32 7.15 536.81 82.72 1.12 0.25 15.94 7.64 Li et al. algorithm 8.83 6.70 695.78 87.14 0.95 0.13 12.08 0.29 PLM 8.89 7.25 524.83 88.92 1.10 0.19 15.81 0.79 LIME 9.86 7.43 719.04 97.80 1.11 0.17 14.51 0.50 Methods 12.47 7.17 1232.96 119.97 1.18 0.24 16.41 0.59

[0127] In summary, combining subjective and objective analysis, we can draw the conclusion that this method can enhance extremely dark images, as well as images with uneven illumination. It can maintain the brightness of bright areas while enhancing dark areas, avoiding over-enhancement, and greatly improve the details, contrast, and clarity of the enhanced images. While the visual effect is good, the objective indicators are also better than the comparison algorithm.

[0128] Example 4

[0129] The embodiment of the present invention provides a low-light image and video enhancement device based on an atmospheric scattering model. Figure 4 , the device comprises:

[0130] The first acquisition module 1 is used to refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance

[0131] The second acquisition module 2 is used to obtain the Transmittance And the superposition coefficient to obtain the final transmittance;

[0132] A global atmospheric light generation module 3 is used to select a preset proportion of pixels from the final transmittance according to the brightness, obtain positions corresponding to the preset proportion of pixels, select V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculate the average value of the V component values ​​as the global atmospheric light;

[0133] A refinement module 4 is used to substitute the final transmittance and the global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, and then refine it to obtain a final enhanced V component;

[0134] The synthesis module 5 is used to keep the hue and saturation components unchanged, convert the image from the HSV' color space to the RGB space using MATLAB to obtain an enhanced image, and finally synthesize the enhanced video.

[0135] Unless otherwise specified, the models of the components in the embodiments of the present invention are not limited, and any device that can perform the above functions may be used.

[0136] The execution subjects of the above modules can be devices with computing functions such as computers, single-chip microcomputers, and microcontrollers. In specific implementation, the embodiments of the present invention do not limit the execution subjects and they are selected according to the needs of actual applications.

[0137] In summary, the images and videos obtained through processing by the above modules in the embodiments of the present invention are natural and have clear details, and can effectively highlight the key information in the images.

[0138] Example 5

[0139] A low-light image and video enhancement device based on atmospheric scattering model, see Figure 5 The device includes: a processor 6 and a memory 7, wherein the memory 6 stores program instructions, and the processor 7 calls the program instructions stored in the memory 6 to enable the device to execute the following method steps in Example 1:

[0140] Refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance

[0141] Based on transmittance Transmittance And the superposition coefficient to obtain the final transmittance;

[0142] Selecting a preset proportion of pixels from the final transmittance according to the brightness, obtaining positions corresponding to the preset proportion of pixels, selecting V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculating the average value of the V component values ​​as the global atmospheric light;

[0143] Substitute the final transmittance and global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, which is then refined to obtain the final enhanced V component;

[0144] Keeping the hue and saturation components unchanged, the image is converted from the HSV' color space to the RGB space using MATLAB to obtain the enhanced image, and finally the enhanced video is synthesized.

[0145] In one embodiment, the V component is refined based on the refinement equation to obtain the transmittance Specifically:

[0146]

[0147] where α is the equilibrium parameter, ||·|| F and ||·|| 1 They refer to the Frobenious norm and l respectively. 1 norm, W is the weight matrix defined, Express First-order derivative filtering is performed in the horizontal and vertical directions.

[0148] In another embodiment, the weight matrix is:

[0149]

[0150]

[0151] Among them, G σ (x, y) is a Gaussian kernel function with standard deviation σ, v is the vertical direction, V(y) is the brightness of point y, W h (x) is the weight in the horizontal direction, W v (x) is the weight in the vertical direction; Ω(x) is the 3×3 neighborhood centered on x; h is the horizontal direction; ε is a constant close to 0 to prevent the denominator from being zero.

[0152] The superposition coefficient is:

[0153]

[0154] Where max(V) is the maximum value of the V component, var(V) is the variance of the V component, and mean(V) is the mean of the V component.

[0155] Furthermore, the initially enhanced V component is obtained, and the final enhanced V component is obtained by refining:

[0156] An S-shaped coefficient is defined using the clarity function to superimpose the high-frequency part representing detail information with the initially enhanced V component;

[0157] Among them, the S-type coefficient function is:

[0158]

[0159] Where k represents the slope of the control function, AG m represents the clarity function;

[0160] The final enhanced V component results are as follows:

[0161]

[0162] Among them, Hig(x,y) represents the high-frequency information of the image, Represents the V component of the initial enhancement.

[0163] In one implementation, MATLAB is used to convert an image from the HSV' color space to the RGB space to obtain an enhanced image, and the enhanced video is finally synthesized as follows:

[0164] Using the hsv2rgb function in MATLAB, the image is converted from the HSV' color space to the RGB space to obtain an enhanced image while keeping the hue and saturation components unchanged;

[0165] Each subsequent frame is enhanced based on the average brightness of the first enhanced frame to ensure the consistency of the average brightness. The enhanced images are then read sequentially and synthesized into an enhanced video at a rate of 30 frames per second.

[0166] It should be pointed out here that the device description in the above embodiment corresponds to the method description in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0167] The execution subjects of the above-mentioned processor 6 and memory 7 can be devices with computing functions such as computers, single-chip microcomputers, and microcontrollers. In specific implementation, the embodiment of the present invention does not limit the execution subject and it is selected according to the needs of actual applications.

[0168] The data signal is transmitted between the memory 7 and the processor 6 via the bus 8, which will not be described in detail in the embodiment of the present invention.

[0169] Example 6

[0170] Based on the same inventive concept, an embodiment of the present invention further provides a computer-readable storage medium, the storage medium includes a stored program, and when the program is running, the device where the storage medium is located is controlled to execute the method steps in the above embodiment.

[0171] The computer-readable storage medium includes but is not limited to a flash memory, a hard disk, a solid-state drive, and the like.

[0172] It should be pointed out here that the description of the readable storage medium in the above embodiment corresponds to the description of the method in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0173] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the present invention are generated.

[0174] The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions may be stored in a computer-readable storage medium or transmitted via a computer-readable storage medium. The computer-readable storage medium may be any available medium that can be accessed by the computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium or a semiconductor medium, etc.

[0175] Those skilled in the art can understand that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0176] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A low-light image and video enhancement method based on an atmospheric scattering model, characterized in that: The method comprises: Refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance Based on transmittance Transmittance And the superposition coefficient to obtain the final transmittance; Selecting a preset proportion of pixels from the final transmittance according to the brightness, obtaining positions corresponding to the preset proportion of pixels, selecting V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculating the average value of the V component values ​​as the global atmospheric light; Substitute the final transmittance and global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, which is then refined to obtain the final enhanced V component; Keeping the hue and saturation components unchanged, the image is converted from the HSV' color space to the RGB space using MATLAB to obtain the enhanced image, and finally the enhanced video is synthesized; The V component is refined based on the refinement equation to obtain the transmittance Specifically: Among them, α is the balance parameter, ∥·∥ F and ∥·∥1 refer to the Frobenious norm and norm, W is the weight matrix defined, Express First-order derivative filtering is performed in the horizontal and vertical directions; The weight matrix is: Among them, G σ (x, y) is a Gaussian kernel function with standard deviation σ, v is the vertical direction, V(y) is the brightness of point y, W h (x) is the weight in the horizontal direction, W v (x) is the weight in the vertical direction; Ω(x) is the 3×3 neighborhood centered on x; h is the horizontal direction; ε is a constant close to 0 to prevent the denominator from being zero.

2. The low-light image and video enhancement method based on the atmospheric scattering model according to claim 1, characterized in that: The superposition coefficient is: Where max(V) is the maximum value of the V component, var(V) is the variance of the V component, and mean(V) is the mean of the V component.

3. The low-light image and video enhancement method based on the atmospheric scattering model according to claim 1, characterized in that: The V component obtained by preliminary enhancement and refined to obtain the final enhanced V component is specifically: An S-shaped coefficient is defined using the clarity function to superimpose the high-frequency part representing detail information with the initially enhanced V component; Among them, the S-type coefficient function is: Where k represents the slope of the control function, AG m represents the clarity function; The final enhanced V component results are as follows: Among them, Hig(x,y) represents the high-frequency information of the image, Represents the V component of the initial enhancement.

4. The low-light image and video enhancement method based on the atmospheric scattering model according to claim 1, characterized in that: The method of using MATLAB to convert the image from the HSV' color space to the RGB space to obtain an enhanced image and finally synthesize the enhanced video is as follows: Using the hsv2rgb function in MATLAB, the image is converted from the HSV' color space to the RGB space to obtain an enhanced image while keeping the hue and saturation components unchanged; Each subsequent frame is enhanced based on the average brightness of the first enhanced frame to ensure the consistency of the average brightness. The enhanced images are then read sequentially and synthesized into an enhanced video at a rate of 30 frames per second.

5. A low-light image and video enhancement device based on an atmospheric scattering model, characterized in that: The device comprises: The first acquisition module is used to refine the V component based on the refinement equation to obtain the transmittance The local maximum value of the V component is obtained to obtain the local maximum map M(x). The local maximum map M(x) is refined using the refinement equation to obtain the transmittance The second acquisition module is used to obtain Transmittance And the superposition coefficient to obtain the final transmittance; Generate a global atmospheric light module, which is used to select a preset proportion of pixels from the final transmittance according to the brightness, obtain the positions corresponding to the preset proportion of pixels, select V component values ​​of the preset proportion of pixels from the low-illuminance image according to these positions, and calculate the average value of the V component value as the global atmospheric light; A refinement module is used to substitute the final transmittance and global atmospheric light into the low-illuminance atmospheric scattering model to obtain a preliminary enhanced V component, and then refine it to obtain the final enhanced V component; A synthesis module is used to keep the hue and saturation components unchanged, convert the image from the HSV' color space to the RGB space using MATLAB to obtain an enhanced image, and finally synthesize the enhanced video; The V component is refined based on the refinement equation to obtain the transmittance Specifically: where α is the equilibrium parameter, ||·|| F and ||·||1 refer to the Frobenious norm and norm, W is the weight matrix defined, Express First-order derivative filtering is performed in the horizontal and vertical directions; The weight matrix is: Among them, G σ (x, y) is a Gaussian kernel function with standard deviation σ, v is the vertical direction, V(y) is the brightness of point y, W h (x) is the weight in the horizontal direction, W v (x) is the weight in the vertical direction; Ω(x) is the 3×3 neighborhood centered on x; h is the horizontal direction; ε is a constant close to 0 to prevent the denominator from being zero.

6. A low-light image and video enhancement device based on an atmospheric scattering model, characterized in that: The device comprises: a processor and a memory, wherein program instructions are stored in the memory, and the processor calls the program instructions stored in the memory to enable the device to execute the method steps described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is enabled to perform the method according to any one of claims 1 to 4.

Images