Underwater non-uniform illumination image restoration method and device based on polarization imaging

Abstract

The invention relates to an underwater non-uniform illumination image recovery method and device based on polarization imaging, and the method integrates non-uniform illumination correction, contrast enhancement and detail improvement, and comprises the following steps: S1, carrying out the preliminary separation of backscattered light and target information light of two polarization sub-images through a low-rank sparse matrix decomposition method; s2, performing non-uniformity correction on the backscattered light from three aspects of adaptive brightness adjustment, gamma correction and local brightness homogenization; s3, fusing the backscattered light subjected to non-uniform correction according to a proportion to obtain homogenized backscattered light; and S4, according to the backscattered light after non-uniform correction, restoring the two polarization sub-images and fusing the two polarization sub-images according to a certain proportion to obtain an underwater clear image with uniform illumination. The non-uniform illumination can be effectively corrected, the overexposure phenomenon is avoided, and the image contrast and definition are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater polarization imaging, and in particular to an underwater non-uniform illumination image restoration method and device based on polarization imaging. Background Art

[0002] When imaging in turbid water, the contrast and clarity of the image are greatly reduced due to the scattering and absorption of light by suspended particles in the water. The restoration method based on the physical model is based on the underwater imaging model. By analyzing the degradation mechanism of the underwater image, the backscattered light and transmittance of the underwater imaging model at infinity are estimated to realize the inversion degradation process, thereby improving the underwater imaging quality. However, due to the limitations of the environment and equipment, active light sources are often required for illumination during underwater night imaging and deep-sea imaging. This non-uniform illumination will introduce over-bright or over-dark areas in the underwater image, making it difficult to estimate the relevant parameters in the underwater imaging model, and easily produce overexposure or underexposure in subsequent image processing, resulting in the loss of target information in the scene. The sharp decline in image quality brings challenges to subsequent image restoration and affects subsequent tasks such as underwater target recognition and target tracking. In view of the impact of non-uniform illumination on underwater image restoration, current processing methods mainly focus on the accurate estimation of backscattered light and the separation of illumination components. However, the non-uniformity of background scattered light is closely related to the non-uniformity of illumination. It is difficult to fully suppress the backscattered light and achieve uniform illumination by simply improving the estimation of backscattered light or extracting the illumination component for operation. Summary of the invention

[0003] The present invention aims to address the deficiencies of the prior art and the problem of image quality degradation under underwater non-uniform illumination conditions, and to provide an underwater non-uniform illumination image restoration method and device based on polarization imaging, which can effectively correct non-uniform illumination, avoid overexposure, and improve image contrast and clarity.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] An underwater non-uniform illumination image restoration method based on polarization imaging combines non-uniform illumination correction, contrast enhancement and detail enhancement, including the following steps:

[0006] S1. Using low-rank sparse matrix decomposition method to preliminarily separate the backscattered light and target information light of the two polarization sub-images;

[0007] S2, non-uniform correction of backscattered light from three aspects: adaptive brightness adjustment, gamma correction, and local brightness uniformity;

[0008] S3, blending the non-uniformly corrected backscattered light in proportion to obtain uniform backscattered light;

[0009] S4. Based on the non-uniformly corrected backscattered light, the two polarization sub-images are restored and fused in a certain ratio to obtain a clear underwater image with uniform illumination.

[0010] The specific steps of step S1 are:

[0011] Three images with polarization angles of 0°, 45°, and 90° are collected. The Stokes vector is used to represent the image with any polarization angle, and two polarization sub-images to be processed are selected according to the image contrast. The low-rank sparse matrix decomposition of the two polarization sub-images is performed by taking advantage of the low-rank property of backscattered light and the sparse property of target information light during underwater imaging, as shown in formula (1):

[0012]

[0013] in,

[0014] ||·|| * is the nuclear norm of the matrix;

[0015] Τr(·) is the trace of the matrix;

[0016] ||·‖ 1 is the matrix l 1 norm;

[0017] G(·) is the forward transform operator, usually the discrete cosine transform;

[0018] The parameters λ and γ are used to balance the sparsity of the low-rank component, the sparse component in the spatial domain, and the low-rank component in the transform domain;

[0019] The augmented Lagrangian function of formula (1) is constructed and optimized by the ADMM algorithm to obtain the backscattered light matrix B with low rank and the target information light matrix D with sparsity.

[0020] In step S2, preprocessing is performed before non-uniform correction. When the image is decomposed by a low-rank sparse matrix, the low-rank matrix may still contain detailed information of some images. From the frequency domain perspective, the image detail information belongs to high-frequency information, and the backscattered light is low-frequency information. The decomposed backscattered light B is smoothed by Gaussian low-pass filtering to filter out the detailed information and obtain a more accurate backscattered light B. 0 .

[0021] In step S2, non-uniform correction is performed on the backscattered light from three aspects, and the specific steps are as follows:

[0022] S21, Adaptive brightness adjustment:

[0023] Backscattered light B after Gaussian filtering0 Perform adaptive brightness adjustment;

[0024] Take the backscattered light B 0 The average value of the pixels in the top 10% of the pixel values ​​is used as the threshold of the overbright pixels, denoted by B light ;

[0025] Take the backscattered light B 0 The average value of the pixels in the last 10% of the pixels is used as the threshold of the dark pixels, denoted by B dark ;

[0026] The adaptive brightness adjustment process is expressed as:

[0027]

[0028] in,

[0029] B 1 (x, y) is the backscattered light image after adaptive brightness adjustment;

[0030] When image B 0 The pixel value of a pixel in is greater than B light When the pixel value is set to B light ; When image B 0 The pixel value of a pixel in is less than B dark When the pixel value is set to B dark ; When image B 0 The pixel value of a certain pixel in [B dark ,B light ] No processing is performed within the interval;

[0031] S22, Gamma Correction:

[0032] Use gamma correction to adjust the brightness of backscattered light from an overall perspective:

[0033]

[0034] in,

[0035] B 2 is the gamma-corrected backscattered light;

[0036] gamma is the gamma correction coefficient;

[0037] For the smoothed backscattered light B 0 Gamma correction can increase the intensity of pixels in dark areas and reduce the intensity of pixels in bright areas, making the image brightness distribution uniform.

[0038] S23, local brightness uniformity:

[0039] Gamma-corrected backscattered light B 2 Perform block processing to enhance the local uniformity of backscattered light; for an image with an image size of m×n, decompose the image into several blocks according to 1 / 10 of the image size. The area is processed, the mean of each area is obtained, and the values ​​are reassigned to all points in the area:

[0040]

[0041] in,

[0042] Ω(x,y) is the local neighborhood of the center point (x,y);

[0043] Δ is the total number of pixels in the Ω(x,y) area;

[0044] The backscattered light after block processing is guided and filtered to eliminate distortion and obtain backscattered light with uniform local brightness.

[0045] In step S3, the backscattered light after adaptive brightness adjustment, gamma correction, and local brightness uniformization is proportionally fused to obtain uniform backscattered light B fusion , the specific formula is:

[0046] B fusion =αB 1 +βB 2 +(1-α-β)B 3 (5);

[0047] Among them, α and β are the proportional coefficients of fusion and are constants between 0 and 1.

[0048] In step S4, the two polarization sub-images are restored according to their features, and the target information light D is decomposed by the low-rank sparse matrix. best To obtain the undegraded image:

[0049]

[0050] Among them, t best For L best The transmittance of

[0051] Estimated image I worst The corresponding transmittance t worst , the undegraded image L worst It can be expressed as:

[0052]

[0053] L bestContains rich image details, L worst The backscattered light distribution is more uniform. best and L worst The specific formula for obtaining the image restoration result L by fusion according to a certain ratio is:

[0054] L=ω·L best +(1-ω)·L worst (8);

[0055] Among them, the fusion parameter ω is a constant between 0 and 1.

[0056] The present invention also provides a restoration device for an underwater non-uniform illumination image restoration method based on polarization imaging, comprising:

[0057] A low-rank sparse matrix decomposition unit is used to preliminarily separate the backscattered light with low rank and the target information light with sparseness;

[0058] A non-uniform correction unit, which is used to correct non-uniform backscattered light, including adaptive brightness adjustment, gamma correction, local brightness uniformization and final fusion operation;

[0059] The image restoration unit is used to restore polarization sub-images with different characteristics respectively and obtain a clear underwater image by fusion.

[0060] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the underwater non-uniform illumination image restoration method based on polarization imaging are implemented.

[0061] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the underwater non-uniform illumination image restoration method based on polarization imaging.

[0062] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0063] 1. The present invention is based on the difference between backscattered light and target information light in matrix dimension. Backscattered light has low rank and target information light has sparseness. The backscattered light and target information light of the image are separated by a low rank sparse matrix decomposition method, thereby achieving preliminary suppression of backscattered light.

[0064] 2. The present invention performs adaptive brightness adjustment on the backscattered light to avoid the influence of too bright or too dark pixels; performs gamma correction to adjust the image brightness from an overall perspective; divides the image into blocks to make the local brightness uniform, and finally fuses the three parts of the corrected backscattered light as the backscattered light for uniform illumination, thereby achieving non-uniform correction of illumination.

[0065] 3. Based on the characteristics of two polarized sub-images, the present invention converts the L containing rich image detail information into best Compared with L, which contains more backscattered light information worst , and are fused in a certain proportion, achieving better results in both non-uniform correction of backscattered light and improvement of image contrast. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of the process of the underwater non-uniform illumination image restoration method based on polarization imaging in the present invention;

[0067] Figure 2 This is a comparison diagram of the descattering results of the present invention and other methods in turbid water bodies of different concentrations;

[0068] Figure 3 for Figure 2 Comparison chart of the effect of partial magnification of the image;

[0069] The following is a detailed description of the embodiments of the present invention with reference to the accompanying drawings. DETAILED DESCRIPTION

[0070] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The embodiments are only used to explain the present invention and are not used to limit the scope of the present invention. The present invention is described in more detail by way of example with reference to the accompanying drawings in the following paragraphs. The advantages and features of the present invention will become clearer according to the following description. It should be noted that the accompanying drawings are all in a very simplified form and are not in precise proportions, and are only used to facilitate and clearly assist in explaining the purpose of the embodiments of the present invention.

[0071] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0072] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0073] In order to verify the effectiveness of the present invention for underwater image enhancement, this embodiment selects underwater images of different scenes as a test set, and makes subjective and objective comparisons with the methods of Schechner, DCP, Chen, Dong, Hu and Wang. Comparison methods 1-6 correspond to the test methods of Schechner, DCP, Chen, Dong, Hu and Wang, respectively.

[0074] like Figure 1 As shown, an underwater non-uniform illumination image restoration method based on polarization imaging combines non-uniform illumination correction, contrast enhancement and detail enhancement, including the following steps:

[0075] Step 1: Use low-rank sparse matrix decomposition method to preliminarily separate the backscattered light and target information light of the two polarization sub-images.

[0076] Step 101: collect three images with polarization angles of 0°, 45°, and 90° respectively, use Stokes vectors to represent images of any polarization angle, and select two polarization sub-images to be processed according to image contrast.

[0077] Step 102: When underwater imaging is performed, the backscattered light has a low rank property and the target information light has a sparse property, and the two polarization sub-images are decomposed into a low rank sparse matrix, as shown in formula (1):

[0078]

[0079] in,

[0080] ‖·|| * is the nuclear norm of the matrix;

[0081] Τr(·) is the trace of the matrix;

[0082] ||·|| 1 is the matrix l 1 norm;

[0083] G(·) is the forward transform operator, usually the discrete cosine transform.

[0084] The parameters λ and γ are used to balance the sparsity of low-rank components, sparse components in the spatial domain, and low-rank components in the transform domain. The augmented Lagrangian function of formula (1) is constructed and optimized by the ADMM algorithm to obtain the backscattered light matrix B with low rank and the target information light matrix D with sparseness.

[0085] When performing low-rank sparse matrix decomposition on an image, the low-rank matrix may still contain some detailed information of the image. From the frequency domain perspective, the image detail information belongs to high-frequency information, and the backscattered light is low-frequency information. The decomposed backscattered light B is smoothed by Gaussian low-pass filtering to filter out the detailed information and obtain a more accurate backscattered light B. 0 .

[0086] Step 2: Perform non-uniform correction on the backscattered light from three aspects: adaptive brightness adjustment, gamma correction, and local brightness uniformity.

[0087] Step 201: Perform non-uniform correction on the backscattered light from three aspects, as follows:

[0088] 1) Adaptive brightness adjustment

[0089] Backscattered light B after Gaussian filtering 0 Perform adaptive brightness adjustment.

[0090] In this embodiment, the backscattered light B 0 The average value of the pixels in the top 10% of the pixel values ​​is used as the threshold of the overbright pixels, denoted by B light Similarly, take the backscattered light B 0 The average value of the pixels in the last 10% of the pixels is used as the threshold of the dark pixels, denoted by B dark The adaptive brightness adjustment process is expressed as:

[0091]

[0092] Among them, B 1 (x, y) is the backscattered light image after adaptive brightness adjustment.

[0093] When image B 0 The pixel value of a pixel in is greater than B light When the pixel value is set to B light ; When image B 0 The pixel value of a pixel in is less than B dark When the pixel value is set to B dark ; When image B 0 The pixel value of a certain pixel in [B dark ,B light ] No processing is performed within the interval;

[0094] 2) Gamma correction

[0095] Use gamma correction to adjust the brightness of backscattered light from an overall perspective:

[0096]

[0097] Among them, B 2 is the backscattered light after gamma correction, and gamma is the gamma correction coefficient. 0 Gamma correction can increase the intensity of pixels in dark areas and reduce the intensity of pixels in bright areas, making the image brightness distribution uniform.

[0098] 3) Local brightness uniformity

[0099] Gamma-corrected backscattered light B 2 Perform block processing to enhance the local uniformity of backscattered light. For an image with an image size of m×n, decompose the image into several blocks according to 1 / 10 of the image size. The area is processed, the mean of each area is obtained, and the values ​​are reassigned to all points in the area:

[0100]

[0101] in,

[0102] Ω(x,y) is the local neighborhood of the center point (x,y);

[0103] Δ is the total number of pixels in the Ω(x,y) region.

[0104] The backscattered light after block processing is guided and filtered to eliminate distortion and obtain backscattered light with uniform local brightness.

[0105] Step 3: The non-uniformly corrected backscattered light is proportionally merged to obtain uniform backscattered light.

[0106] Step 301: The backscattered light after adaptive brightness adjustment, gamma correction, and local brightness uniformization is proportionally fused to obtain uniform backscattered light B. fusion , the specific formula is:

[0107] B fusion =αB 1 +βB 2 +(1-α-β)B 3 (5);

[0108] Among them, α and β are the proportional coefficients of fusion and are constants between 0 and 1.

[0109] In this embodiment, during the experiment, α is set to 0.1, β is set to 0.4, and the image is processed.

[0110] Step 4: Based on the non-uniformly corrected backscattered light, the two polarization sub-images are restored and fused in a certain ratio to obtain a clear underwater image with uniform illumination.

[0111] The target information light D is decomposed by the low-rank sparse matrix. best Contains more target information, so according to D best To obtain the undegraded image:

[0112]

[0113] Among them, t best For L best The transmittance of

[0114] Estimated image I worst The corresponding transmittance t worst , the undegraded image L worst It can be expressed as:

[0115]

[0116] L best Contains rich image details, L worst The backscattered light distribution is more uniform. best and L worst The specific formula for obtaining the image restoration result L by fusion according to a certain ratio is:

[0117] L=ω·L best +(1-ω)·L worst (8);

[0118] The fusion parameter ω is a constant between 0 and 1. A suitable parameter can be selected for processing according to the scene naturalness and image contrast of the image. In this embodiment, ω is taken as 0.8 for processing to obtain the final underwater enhanced image L.

[0119] like Figure 2 The following is a comparison of the enhancement effects of this embodiment and other six methods on images under non-uniform illumination of turbid water bodies of different concentrations. Experiments were conducted in turbid water bodies of low, medium and high concentrations, and some of the experimental results are selected as follows: Figure 2As shown, the image data rows 1-2 correspond to low turbidity water imaging, the image data rows 3-4 correspond to medium turbidity water imaging, and the image data rows 5-6 correspond to high turbidity water imaging. It can be seen that compared with other methods, the overall illumination distribution of the image processed by the method of the present invention is more uniform, the brightness of the darker areas in the original image is improved, and the brightness of the brighter areas is suppressed, there is no obvious overexposure phenomenon and the visual effect is better. In terms of image enhancement, the image clarity and contrast are greatly improved, and the image detail information is enhanced.

[0120] In order to better display the details of the image, Figure 3 The detailed comparison diagrams of descattering results in turbid water bodies of different concentrations are sorted out, showing the local magnified views of the processing results of the rectangular area marked in the original image by different methods in low, medium and high turbid water bodies. From the local magnified view, it can be found that the algorithm of the present invention can well restore the target information submerged in the scattering. As the turbidity of the water body increases, the target information of the original image is almost indistinguishable, while the image processed by this algorithm can still obtain good clarity and contrast. It can be seen that in turbid water bodies of different concentrations, the method of the present invention is more superior than the processing results of other methods, and the image detail information is richer, and as the turbidity of the water body increases, the superiority is more obvious.

[0121] This embodiment uses five image quality evaluation indicators: image eight-neighborhood contrast, EME (enhancement measure evaluation), AG (Average Gradient), EI (Edge Intensity), and UIQM (Underwater Image Quality Measures) to evaluate the image quality. Figure 2 The experimental results were quantitatively analyzed.

[0122] Eight-neighborhood contrast is used to measure the richness of image detail information. The larger the value, the richer the image details.

[0123] EME is an image enhancement evaluation index. It divides the image into several block areas and takes the average of the logarithm of the ratio of the maximum and minimum grayscale values ​​of the image in each block area to measure the degree of local grayscale change in the image. The larger the EME value, the more drastic the image grayscale change and the stronger the image details.

[0124] AG refers to the average gradient of the image. The larger its value is, the higher the image clarity is.

[0125] EI characterizes the clarity of image edge information by extracting the gradient values ​​in the horizontal and vertical directions of the image. The larger the EI value, the more obvious the image edge information.

[0126] UIQM is an underwater image quality evaluation, which is a linear combination of the color measurement index (UICM), the clarity measurement index (UISM), and the contrast measurement index (UIConM). In this embodiment, the influence of color on the UIQM value is ignored, and the relative size of the comparison result is used to characterize the image quality. The larger the UIQM value, the clearer the image and the higher the contrast.

[0127] Table 1 summarizes the average values ​​of the evaluation indicators of each group of image data. The optimal values ​​in the results of each image quality evaluation indicator are marked in bold, and the suboptimal values ​​are marked with underlines. From the data in Table 1, the method proposed in this embodiment is the optimal value or suboptimal value in the five image evaluation indicators. Compared with other methods, it is stable in turbid water bodies of different concentrations and has a good descattering effect. From a general point of view, this also proves that the method of the present invention can not only effectively reduce the impact of non-uniform illumination, but also greatly improve the image contrast and edge information richness, and also has a good effect in improving image detail information and clarity, which can significantly improve image quality.

[0128] Table 1 Image quality evaluation results

[0129]

[0130] The present invention also provides an underwater non-uniform illumination image restoration method and device based on polarization imaging, comprising:

[0131] A low-rank sparse matrix decomposition unit is used to preliminarily separate the backscattered light with low rank and the target information light with sparseness;

[0132] A non-uniform correction unit, which is used to correct non-uniform backscattered light, including adaptive brightness adjustment, gamma correction, local brightness uniformization and final fusion operation;

[0133] The image restoration unit is used to restore polarization sub-images with different characteristics respectively and obtain a clear underwater image by fusion.

[0134] The present invention also provides a specific implementation of an electronic device capable of implementing all steps in the underwater non-uniform illumination image restoration method based on polarization imaging in the above embodiment, and the electronic device specifically includes the following contents:

[0135] Processor, memory, communications interface and bus;

[0136] Among them, the processor, memory, and communication interface communicate with each other through the bus; the communication interface is used to realize information transmission between related devices such as server-side devices, metering devices, and user-side devices.

[0137] The processor is used to call the computer program in the memory, and when the processor executes the computer program, all the steps in the underwater image enhancement method in the above embodiment are implemented.

[0138] The present invention also provides a computer-readable storage medium capable of implementing all the steps of the underwater non-uniform illumination image restoration method based on polarization imaging in the above-mentioned embodiment. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, all the steps of the underwater non-uniform illumination image restoration method based on polarization imaging in the above-mentioned embodiment are implemented.

[0139] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the hardware + program embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0140] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0141] Although the present application provides method operation steps such as embodiments or flow charts, more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in the embodiments is only one way of executing the order of many steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in the order of the method shown in the embodiments or the drawings or in parallel (for example, in a parallel processor or multi-threaded processing environment).

[0142] Those skilled in the art will appreciate 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. Moreover, 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.

[0143] 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 comprising 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.

[0144] 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. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0145] The present invention is described above by way of example in conjunction with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as various improvements are made using the method concept and technical solution of the present invention, or are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. An underwater non-uniform illumination image restoration method based on polarization imaging, which combines non-uniform illumination correction, contrast enhancement and detail enhancement, is characterized in that: The following steps are involved: S1. Using low-rank sparse matrix decomposition method to preliminarily separate the backscattered light and target information light of the two polarization sub-images; S2, non-uniform correction of backscattered light from three aspects: adaptive brightness adjustment, gamma correction, and local brightness uniformity; S3, blending the non-uniformly corrected backscattered light in proportion to obtain uniform backscattered light; S4. Based on the non-uniformly corrected backscattered light, the two polarization sub-images are restored and fused in a certain ratio to obtain a clear underwater image with uniform illumination.

2. According to claim 1, the method for restoring underwater non-uniform illumination images based on polarization imaging is characterized in that: The specific steps of step S1 are: Three images with polarization angles of 0°, 45°, and 90° are collected. The Stokes vector is used to represent the image with any polarization angle, and two polarization sub-images to be processed are selected according to the image contrast. The low-rank sparse matrix decomposition of the two polarization sub-images is performed by taking advantage of the low-rank property of backscattered light and the sparse property of target information light during underwater imaging, as shown in formula (1): in, ||·|| * is the nuclear norm of the matrix; Τr(·) is the trace of the matrix; ||·||1 is the l1 norm of the matrix; G(·) is the forward transformation operator; The parameters λ and γ are used to balance the sparsity of the low-rank component, the sparse component in the spatial domain, and the low-rank component in the transform domain; The augmented Lagrangian function of formula (1) is constructed and optimized by the ADMM algorithm to obtain the backscattered light matrix B with low rank and the target information light matrix D with sparsity.

3. The underwater non-uniform illumination image restoration method based on polarization imaging according to claim 1, characterized in that: In step S2, preprocessing is performed before non-uniform correction. When the image is decomposed into a low-rank sparse matrix, the low-rank matrix also contains detailed information of some images. From the frequency domain perspective, the image detail information belongs to high-frequency information, and the backscattered light is low-frequency information. The decomposed backscattered light B is smoothed by Gaussian low-pass filtering to filter out the detailed information and obtain more accurate backscattered light B0.

4. The underwater non-uniform illumination image restoration method based on polarization imaging according to claim 1, characterized in that: In step S2, non-uniform correction is performed on the backscattered light from three aspects, and the specific steps are as follows: S21, Adaptive brightness adjustment: Adaptively adjust the brightness of the backscattered light B0 after Gaussian filtering and smoothing; The average value of the top 10% of the pixels in the backscattered light B0 is taken as the threshold of the overbright pixels, denoted as B light ; The average value of the pixels in the last 10% of the backscattered light B0 is taken as the threshold of the dark pixels, denoted as B dark ; The adaptive brightness adjustment process is expressed as: in, B1(x,y) is the backscattered light image after adaptive brightness adjustment; When the pixel value of a pixel in image B0 is greater than B light When the pixel value is set to B light ; When the pixel value of a pixel in image B0 is less than B dark When the pixel value is set to B dark ; When the pixel value of a pixel in image B0 is between [B dark ,B light ] No processing is performed within the interval; S22, Gamma Correction: Use gamma correction to adjust the brightness of backscattered light from an overall perspective: in, B2 is the backscattered light after gamma correction; gamma is the gamma correction coefficient; Gamma correction is performed on the smoothed backscattered light B0 to increase the intensity value of the pixels in the dark area and reduce the intensity value of the pixels in the overly bright area, so that the image brightness distribution tends to be uniform; S23, local brightness uniformity: The gamma-corrected backscattered light B2 is processed in blocks to enhance the local uniformity of the backscattered light. For an image with an image size of m×n, the image is decomposed into several blocks according to 1 / 10 of the image size. The area is processed, the mean of each area is obtained, and the values ​​are reassigned to all points in the area: in, Ω(x,y) is the local neighborhood of the center point (x,y); Δ is the total number of pixels in the Ω(x,y) area; The backscattered light after block processing is guided and filtered to eliminate distortion and obtain backscattered light with uniform local brightness.

5. The underwater non-uniform illumination image restoration method based on polarization imaging according to claim 1, characterized in that: In step S3, the backscattered light after adaptive brightness adjustment, gamma correction, and local brightness uniformization is proportionally fused to obtain uniform backscattered light B fusion , the specific formula is: B fusion =αB1+βB2+(1-α-β)B3 (5); Among them, α and β are the proportional coefficients of fusion and are constants between 0 and 1.

6. The underwater non-uniform illumination image restoration method based on polarization imaging according to claim 1, characterized in that: In step S4, the two polarization sub-images are restored according to their features, and the target information light D is decomposed by the low-rank sparse matrix. best To obtain the undegraded image: Among them, t best For L best The transmittance of Estimated image I worst The corresponding transmittance t worst , the undegraded image L worst It can be expressed as: L best and L worst The specific formula for obtaining the image restoration result L by fusion according to a certain ratio is: L=ω·L best +(1-ω)·L worst (8); Among them, the fusion parameter ω is a constant between 0 and 1.

7. A restoration device for an underwater non-uniform illumination image restoration method based on polarization imaging according to any one of claims 1 to 6, characterized in that: include: A low-rank sparse matrix decomposition unit is used to preliminarily separate the backscattered light with low rank and the target information light with sparseness; A non-uniform correction unit, which is used to correct non-uniform backscattered light, including adaptive brightness adjustment, gamma correction, local brightness uniformization and final fusion operation; The image restoration unit is used to restore polarization sub-images with different characteristics respectively and obtain a clear underwater image by fusion.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the underwater non-uniform illumination image restoration method based on polarization imaging described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the underwater non-uniform illumination image restoration method based on polarization imaging as described in any one of claims 1 to 6 are implemented.

Images