An underwater image enhancement method and system

Abstract

The present invention discloses an underwater image enhancement method and system, including: performing L-channel contrast enhancement on the original underwater image to obtain a locally enhanced image; using the contrast-limited adaptive histogram equalization method to perform color restoration on the original underwater image to obtain an equalized image; linearly weighted fusing the locally enhanced image and the equalized image to obtain a fused image; performing color correction on the fused image for the #imgabs0# channel and the #imgabs1# channel to obtain a corrected image; performing depth prediction on the fused image through a monocular depth estimation method to obtain a depth guidance map; performing RGB channel information expansion on the fused image to obtain a detail image; and fusing the corrected image, the depth guidance map, and the detail image to obtain a final enhanced image. The present invention realizes the coordinated fusion of local contrast enhancement and global color restoration, so that the visual quality of the enhanced image is significantly improved.

Description

Technical Field

[0001] The present invention belongs to the field of image enhancement, and particularly relates to an underwater image enhancement method and system. Background Art

[0002] Underwater machine vision is like the human eye, helping humans see clearly the underwater world. There are many adverse factors that are difficult to solve at present in the ocean scene, such as high turbidity, single color, complex underwater background, uneven illumination, etc., which make the images captured by underwater imaging systems generally have problems such as low resolution, low contrast, blurred texture, color distortion, and quality degradation.

[0003] To solve the problems existing in underwater images, in recent years, researchers have proposed many underwater image enhancement methods to solve this problem. Local contrast enhancement methods such as adaptive histogram equalization can improve the details in the dark areas, but are prone to cause global color imbalance, making the image color deviate from nature. And global color restoration methods such as subtraction map-guided fusion may weaken the local area contrast when correcting the color, making the details blurred. Therefore, it is difficult to coordinate local contrast enhancement and global color restoration in traditional methods, seriously affecting the effect of underwater image enhancement. Summary of the Invention

[0004] The present invention provides an underwater image enhancement method and system, which effectively overcomes the technical limitation that it is difficult to coordinate local contrast enhancement and global color restoration in traditional methods through a collaborative mechanism of brightness-color separation processing and depth-guided correction.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] The first aspect of the present invention provides an underwater image enhancement method, including:

[0007] Obtain the original underwater image, perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image; use the limited contrast adaptive histogram equalization method to perform color restoration on the original underwater image to obtain an equalized image;

[0008] Linearly weighted fuse the locally enhanced image and the equalized image to obtain a fused image;

[0009] Perform channel and channel color correction on the fused image in the CIELab color space to obtain a corrected image; perform depth prediction on the fused image through a monocular depth estimation method to obtain a depth guidance map; perform RGB channel information expansion on the fused image to obtain a detail image;

[0010] Fuse the corrected image, the depth guidance map and the detail image to obtain a final enhanced image.

[0011] Further, the L-channel contrast of the original underwater image is enhanced in the CIELab color space to obtain a locally enhanced image, and the expression formula is:

[0012] The L-channel enhanced image feature is obtained by enhancing the L-channel contrast of the original underwater image in the CIELab color space, and the expression formula is:

[0013]

[0014]

[0015] In the formula, is the average value of the local pixel block; represents the number of pixel points in the local pixel block; r is the window radius of the mean filter; represents the pixel value at the coordinate in the original underwater image in the L channel; is the L-channel enhanced image feature; is the pixel value at the coordinate in the original underwater image in the L channel; n and m are the serial numbers of the pixel points in the local pixel block;

[0016] The original underwater image is sent into the CIELab color space channel and channel to obtain the red-green chromaticity feature and the yellow-cyan chromaticity feature;

[0017] The L-channel enhanced image feature, the red-green chromaticity feature, and the yellow-cyan chromaticity feature are superimposed to obtain a locally enhanced image.

[0018] Further, the color of the original underwater image is restored by using the contrast-limited adaptive histogram equalization method to obtain an equalized image, specifically including:

[0019] The original underwater image is decomposed into three-channel initial images; the three-channel initial images are the R-channel initial image, the G-channel initial image, and the B-channel initial image, respectively;

[0020] After the channel initial images are grayscaled, the channel initial images are divided into M×N sub-image blocks, and the histograms of the sub-image blocks are calculated respectively, and the expression formula is:

[0021]

[0022]

[0023]

[0024] In the formula, are the coordinates after grayscale conversion of the initial channel image of the pixels; is the gray level, ; is the Kronecker function, and PS is the set of pixel point coordinates of the initial channel image; is the frequency of pixels with gray level k in the initial channel image; is the histogram of the sub-image block;

[0025] Perform a clipping operation on the histogram of each sub-image block according to the set clipping threshold to obtain the clipped histogram. The expression formula is:

[0026]

[0027]

[0028]

[0029]

[0030] In the formula, is the clipping threshold;<{ is the clipping coefficient; is the number of gray levels of the sub-image block; is the total number of pixel values exceeding the clipping threshold in the sub-image block; is the average distributed pixel value exceeding the clipping threshold in the sub-image block; is the clipped histogram;

[0031] Perform bilinear interpolation reconstruction on the clipped histogram to obtain the equalized image.

[0032] Furthermore, linearly weight and fuse the locally enhanced image and the equalized image to obtain the fused image, specifically including:

[0033] Calculate the covariance matrix of the locally enhanced image. The expression formula is:

[0034]

[0035]

[0036] Calculate the covariance matrix of the equalized image. The expression formula is:

[0037]

[0038]

[0039] In the formula, is the covariance matrix of the locally enhanced image; is the covariance matrix of the equalized image; To locally enhance the image; To equalize the image; is the average pixel value of the local enhanced image; is the average pixel value of the equalized image; t is the index of the pixel point;

[0040] The eigenvectors are calculated based on the covariance matrix of the local enhanced image and the equalized image, and the principal component space is constructed by the eigenvectors of the local enhanced image and the equalized image; the expression formula is:

[0041] ; ;

[0042] In the formula, is the covariance matrix The characteristic value of is the covariance matrix The eigenvector of is the covariance matrix The characteristic value of is the covariance matrix The eigenvector of is the principal component space corresponding to the local enhanced image; is the principal component space corresponding to the equalized image; 、 、 are the eigenvectors of the first principal component, the second principal component, and the third principal component in the local enhanced image respectively; 、 、 are the eigenvectors of the first principal component, the second principal component, and the third principal component in the equalized image respectively;

[0043] Map the local enhanced image and the equalized image to the principal component space to obtain the key features, which are expressed as follows:

[0044]

[0045] In the formula, Mapping the local enhanced image to the principal component space Obtain key features; Mapping the equalized image to the principal component space Obtain key features; To locally enhance the image; To equalize the image;

[0046] The fusion weight is obtained by calculating the key features; the local enhanced image and the equalized image are fused according to the fusion weight to obtain a fused image.

[0047] Further, fusion weights are calculated from the key features; the local enhanced image and the equalized image are fused according to the fusion weights to obtain a fused image, specifically including:

[0048]

[0049]

[0050] In the formula, is the fusion weight of the local enhanced image; is the fusion weight of the equalized image; t is the index of the pixel point; is the main component serial number; is the fused image.

[0051] Further, color correction is performed on the fused image in the CIELab color space for the channel and the channel to obtain a corrected image, specifically including:

[0052]

[0053]

[0054]

[0055] In the formula, is the corrected image feature of the fused image in the channel, is the image feature of the fused image in the channel, is the average value of the image features of the fused image in the channel; is the corrected image feature of the fused image in the channel, is the image feature of the fused image in the channel, is the average value of the image features of the fused image in the channel; is the corrected image; is the mapping function from the CIELab color space to the RGB color space; is the image feature of the fused image in the channel; c is the channel serial number in the RGB color space, R is the red channel in the RGB color space, G is the green channel in the RGB color space, and B is the green channel in the RGB color space.

[0056] Further, the corrected image, the depth guidance map, and the detail image are fused to obtain a final enhanced image, specifically including:

[0057]

[0058]

[0059] In the formula, is the final enhanced image; is the depth guidance map; is the corrected image; is the image feature of image fusion in the three RGB channels; is the detail image; represents the Gaussian blur processing function.

[0060] The second aspect of the present invention provides an underwater image enhancement system, including:

[0061] An acquisition module for acquiring an original underwater image;

[0062] A local enhancement module for performing L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image;

[0063] An image equalization module for performing color restoration on the original underwater image by using the method of limited contrast adaptive histogram equalization to obtain an equalized image;

[0064] A fusion module for linearly weighted fusion of the locally enhanced image and the equalized image to obtain a fused image;

[0065] A correction module for performing channel and channel color correction on the fused image in the CIELab color space to obtain a corrected image; performing depth prediction on the fused image by using a monocular depth estimation method to obtain a depth guidance map; performing RGB channel information expansion on the fused image to obtain a detail image;

[0066] An output module for performing image fusion on the corrected image, the depth guidance map and the detail image to obtain a final enhanced image.

[0067] The third aspect of the present invention provides an electronic terminal, including a processor and a storage medium; the storage medium is used for storing instructions; the processor is used for operating according to the instructions to execute the steps of the underwater image enhancement method described in the first aspect of the present invention.

[0068] The fourth aspect of the present invention provides an electronic device including a storage medium and a processor; the storage medium is used for storing instructions; the processor is used for operating according to the instructions to execute the underwater image enhancement method described in the first aspect of the present invention.

[0069] Compared with the prior art, the beneficial effects of the present invention:

[0070] By enhancing the contrast of the L channel of the original underwater image in the CIELab color space, the present invention can effectively improve the brightness and contrast of the image, making the originally blurred and dim underwater scene clearer and brighter. Using the contrast-limited adaptive histogram equalization to perform color restoration on the original underwater image can enhance the color saturation and contrast of the image, making the colors more rich and accurate, and correcting the common color cast problem of underwater images.

[0071] The present invention linearly weights and fuses the locally enhanced image and the equalized image. On the one hand, it retains the clear details after local enhancement, and on the other hand, it uses the rich color information after equalization to make the fused image achieve a better balance in terms of brightness, contrast, and color performance, avoiding the deficiencies that may occur in a single processing method.

[0072] By performing color correction on the a channel and b channel of the fused image in the CIELab color space, the present invention can effectively improve the color accuracy of the image, making the colors more real and natural. Using the depth guidance map obtained by monocular depth estimation adds depth information to the image, enhancing the sense of three-dimensionality and spatial hierarchy. The detail image obtained by expanding the RGB channel information of the fused image further enriches the details and textures of the image, making the minute features more clearly visible. Finally, the corrected image, the depth guidance map, and the detail image are fused, integrating multi-dimensional information such as color, depth, and details, so that the enhanced image has been significantly improved in visual quality and can more realistically and comprehensively display the underwater scene. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 is a flowchart of an underwater image enhancement method provided by the present invention;

[0074] Figure 2 is the depth guidance map provided by the present invention;

[0075] Figure 3 is the enhancement result of various methods provided by the present invention on the UCCS dataset;

[0076] Figure 4 is the enhancement result of various methods provided by the present invention on the UIEB dataset;

[0077] Figure 5 is the enhancement result of various methods provided by the present invention on the UIQS dataset. DETAILED DESCRIPTION OF THE INVENTION

[0078] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0079] Example 1

[0080] As Figure 1 shown, this example provides an underwater image enhancement method, including the following steps:

[0081] Obtain the original underwater image, map the original underwater image from the RGB color space to the CIELab color space, and perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image, specifically including:

[0082] Perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain L-channel enhanced image features, and the expression formula is:

[0083]

[0084]

[0085] In the formula, is the average value of the local pixel block; represents the number of pixel points in the local pixel block; r is the window radius of the mean filter; represents the coordinate in the original underwater image in the L-channel of the pixel value; is the L-channel enhanced image feature; is the coordinate in the original underwater image in the L-channel of the pixel value; n and m are the serial numbers of pixel points in the local pixel block;

[0086] Send the original underwater image into the CIELab color space channel and channel to obtain red-green chromaticity features and yellow-cyan chromaticity features; [[ID=4�]]

[0087] Overlay the L-channel enhanced image features, red-green chromaticity features and yellow-cyan chromaticity features to obtain a locally enhanced image.

[0088] This example uses the local contrast method to solve the problems of low contrast, blurring and detail loss of underwater images caused by light scattering and absorption, and enhances the local details of the brightness of underwater images. By enhancing the contrast of the L-channel, the edge and contour information of the underwater image can be highlighted, the three-dimensional sense and layering of the image can be enhanced, and the depth and distance information of the underwater scene can be made more intuitive, which helps to better understand the spatial structure of the underwater environment.

[0089] Use the limited contrast adaptive histogram equalization method to perform color restoration on the original underwater image to obtain an equalized image, specifically including:

[0090] Decompose the original underwater image into three initial channel images; the three initial channel images are the initial R-channel image, the initial G-channel image, and the initial B-channel image respectively;

[0091] After grayscale processing the initial channel images, divide the initial channel images into M×N sub-image blocks, and calculate the histograms of the sub-image blocks respectively. The expression formula is:

[0092]

[0093]

[0094]

[0095] In the formula, is the pixel at the coordinate after grayscale processing of the initial channel image; is the gray level, ; is the Kronecker function, and PS is the set of pixel point coordinates of the initial channel image; is the frequency of the pixel point with gray level k in the initial channel image; is the histogram of the sub-image block; c is the channel number in the RGB color space, R is the red channel in the RGB color space, G is the green channel in the RGB color space, and B is the green channel in the RGB color space;

[0096] Perform a clipping operation on the histogram of each sub-image block according to the set clipping threshold to obtain a clipped histogram. The expression formula is:

[0097]

[0098]

[0099]

[0100]

[0101] In the formula, is the clipping threshold; is the clipping coefficient; is the number of gray levels of the sub-image block; is the total number of pixel values exceeding the clipping threshold in the sub-image block; is the evenly distributed pixel value exceeding the clipping threshold in the sub-image block; is the clipped histogram;

[0102] Perform bilinear interpolation reconstruction on the clipped histogram to obtain an equalized image.

[0103] In this embodiment, local histogram equalization is used to solve the problem of uneven pixel distribution in the image caused by the attenuation of underwater light propagation, stretch the dark details of the red channel, and restore the lost red information due to attenuation. At the same time, the green and blue channels are enhanced to correct some color deviations, making the color distribution of the entire image more uniform and improving the visibility of details.

[0104] The locally enhanced image and the equalized image are linearly weighted and fused to obtain a fused image, specifically including:

[0105] Calculate the covariance matrix of the locally enhanced image, and the expression formula is:

[0106]

[0107]

[0108] Calculate the covariance matrix of the equalized image, and the expression formula is:

[0109]

[0110]

[0111] In the formula, is the covariance matrix of the locally enhanced image; is the covariance matrix of the equalized image; is the locally enhanced image; is the equalized image; is the average pixel value of the locally enhanced image; is the average pixel value of the equalized image; t is the index of the pixel point;

[0112] Calculate the eigenvectors according to the covariance matrices of the locally enhanced image and the equalized image, and construct the principal component space from the eigenvectors of the locally enhanced image and the equalized image; the expression formula is:

[0113] ; ;

[0114] In the formula, is the covariance matrix is the eigenvalue of is the covariance matrix is the eigenvector of is the covariance matrix is the eigenvalue of is the covariance matrix is the eigenvector of is the principal component space corresponding to the locally enhanced image; is the principal component space corresponding to the equalized image; 、 , are the eigenvectors of the first, second, and third principal components in the locally enhanced image, respectively; , , are the eigenvectors of the first, second, and third principal components in the equalized image, respectively;

[0115] The locally enhanced image and the equalized image are mapped to the principal component space to obtain key features. The expression formula is:

[0116]

[0117] In the formula, is the locally enhanced image mapped to the principal component space to obtain key features; is the equalized image mapped to the principal component space to obtain key features;

[0118] The fusion weights are calculated from the key features; the locally enhanced image and the equalized image are fused according to the fusion weights to obtain a fused image. The expression formula is:

[0119]

[0120]

[0121] In the formula, is the fusion weight of the locally enhanced image; is the fusion weight of the equalized image; t is the index of the pixel point; is the principal component serial number; is the fused image.

[0122] The fused image in this embodiment is more excellent than the locally enhanced image in terms of global contrast, and is clearer than the equalized image in terms of local details. While retaining details, it avoids over-enhancement and improves the quality of underwater images. It realizes the coordinated fusion of local contrast enhancement and global color restoration, making the enhanced image significantly improved in visual quality.

[0123] Performing channel and channel color correction on the fused image in the CIELab color space to obtain a corrected image, specifically including:

[0124]

[0125]

[0126]

[0127] In the formula, To fuse the image Corrected image features under channels, To fuse the image Image features under the channel, To fuse the image The average value of image features under the channel; To fuse the image Corrected image features under channels, To fuse the image Image features under the channel, To fuse the image The average value of image features under the channel; To correct the image; is the mapping function from CIELab color space to RGB color space; To fuse the image Image features under channels.

[0128] Performing a-channel and b-channel color correction on the fused image in the CIELab color space can accurately adjust the color components of the image, further optimize color accuracy and consistency, effectively improve the color shift problem of underwater imaging, and make the image color closer to the real scene.

[0129] The depth guidance map is obtained by performing depth prediction on the fused image through the monocular depth estimation method, which introduces the depth information dimension into the image. Figure 2 In the [1], DGM (1) to DGM (3) are the depth guidance maps corresponding to the three images, which help enhance the stereoscopic and spatial sense of the image, enabling the observer to better understand the three-dimensional structure of the underwater scene and the spatial relationship of objects. Expanding the RGB channel information of the fused image to obtain a detail image can mine and enhance the subtle features and edge information that were originally difficult to detect in the image, making the image details richer and more complete, and improving the image resolution and quality. This is of great significance for observing the tiny features of underwater organisms and subtle changes in terrain.

[0130] The rectified image, the depth guidance map, and the detail image are fused to obtain the final enhanced image, specifically including:

[0131]

[0132]

[0133] In the formula, To enhance the final image; is the depth guidance map; To correct the image; are the image features of the image fusion in the RGB three channels; is the detail image; is expressed as the Gaussian blur processing function.

[0134] To fully verify the enhancement effect of the method of this embodiment on underwater images, comprehensive quantitative and qualitative evaluations were carried out on three standard datasets: the UCCS dataset, the UIQS dataset, and the UIEB dataset. The UCCS dataset includes: the blue color subset, the green color subset, and the blue-green color subset, which are used to verify the color correction ability of the method of this embodiment. UIQS includes subset, subset, subset, subset, and subset, where each subset has 726 pictures, which are used to verify the visibility ability of the method of this embodiment. The UIEB dataset includes underwater images with different scenes and different degradation degrees, which are used to verify the generalization performance of the method of this embodiment.

[0135] The method of this embodiment was compared with 10 existing methods. The 10 existing methods include: IBLA (underwater image restoration method based on image blur and light absorption), GDCP (generalized method of dark channel prior), HFM (hybrid fusion method), ULAP (method of image restoration using underwater light attenuation prior), WCBF (weighted color balance and fusion method), PCFB (foreground and background principal component fusion method), CLUIE-Net (underwater image enhancement network model based on comparative learning), LANet (adaptive learning attention network model), PUIE-Net (probabilistic underwater image enhancement network), and HCLR-Net (contrast learning regularization network combined with local random perturbation).

[0136] The evaluation metrics of the method of this embodiment and the 10 existing methods include EI (edge intensity), IE (information entropy), CCF (color contrast fog density index), and AG (average gradient); the intensity of edge information in the image is measured by the edge intensity; the amount of information in the image is measured by the information entropy; the color restoration degree and clarity of the image are evaluated by the color contrast fog density index; the average gradient measures the severity of the change in pixel values in the image.

[0137] Table 1, comparison table of the quantitative evaluation experimental results of the method of this embodiment and the 10 existing methods on the UCCS dataset;

[0138]

[0139] Table 2. Comparison table of quantitative evaluation experimental results between the method of this embodiment and 10 existing methods on the UIQS dataset;

[0140]

[0141] Table 3. Comparison table of quantitative evaluation experimental results between the method of this embodiment and 10 existing methods on the UIEB dataset;

[0142]

[0143] In Figures 3 to 5 : (a) is the original underwater image, (b) to (k) are the enhanced images output by IBLA, GDCP, HFM, ULAP, WCBF, PCFB, CLUIE-Net, LANet, PUIE-Net, HCLR-Net, and ( ) is the enhanced image output by this embodiment. By comparing the enhancement results, it can be judged that: the method of this embodiment is not only more natural in color restoration, but also has better performance in detail preservation and noise suppression. Combining Tables 1 to 3, it can be seen that this embodiment can reduce noise interference while enhancing the contrast, and the depth map-guided color correction method can more accurately adjust the color distribution, making the finally generated image superior to the existing methods in both visual perception and objective evaluation indicators.

[0144] Embodiment 2

[0145] This embodiment also provides an underwater image enhancement system, which can execute the underwater image enhancement method described in Embodiment 1. The underwater image enhancement system includes:

[0146] An acquisition module for acquiring the original underwater image;

[0147] A local enhancement module for performing L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image;

[0148] An image equalization module for performing color restoration on the original underwater image using the contrast-limited adaptive histogram equalization method to obtain an equalized image;

[0149] A fusion module for linearly weighted fusion of the locally enhanced image and the equalized image to obtain a fused image;

[0150] A correction module for performing channel and channel color correction on the fused image in the CIELab color space to obtain a corrected image; performing depth prediction on the fused image through a monocular depth estimation method to obtain a depth guidance map; and performing RGB channel information expansion on the fused image to obtain a detail image;

[0151] An output module, configured to perform image fusion on the corrected image, the depth guidance map, and the detail image to obtain a final enhanced image.

[0152] Embodiment 3

[0153] This embodiment also provides an electronic terminal, including a processor and a storage medium; the storage medium is used for storing instructions; the processor is configured to operate according to the instructions to execute the steps of the underwater image enhancement method described in Embodiment 1.

[0154] Embodiment 4

[0155] This embodiment also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the underwater image enhancement method described in Embodiment 1 are implemented.

[0156] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt 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.

[0157] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows 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 the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 one or more processes and / or blocks Figure 1 one or more blocks the device with the functions specified in the block or blocks.

[0158] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements in the process Figure 1 one or more processes and / or blocks Figure 1 one or more blocks the functions specified in the block or blocks.

[0159] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the steps specified in one process or multiple processes and / or blocks Figure 1 One process or multiple processes and / or blocks Figure 1 Steps for implementing the functions specified in one block or multiple blocks

[0160] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. An underwater image enhancement method, characterized in that, Including: Obtain the original underwater image, perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image; use the Contrast Limited Adaptive Histogram Equalization method to perform color restoration on the original underwater image to obtain an equalized image; Linearly weighted fuse the locally enhanced image and the equalized image to obtain a fused image; Perform the following operations on the fused image in the CIELab color space Channel and Channel color correction to obtain a corrected image; perform depth prediction on the fused image through a monocular depth estimation method to obtain a depth guidance map; perform RGB channel information expansion on the fused image to obtain a detail image; Perform image fusion on the corrected image, depth guidance map, and detail image to obtain the final enhanced image.

2. The underwater image enhancement method according to claim 1, wherein Perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image, and the expression formula is: Perform L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain L-channel enhanced image features, and the expression formula is: ； ； In the formula, is the average value of the local pixel block; represents the number of pixel points in the local pixel block; r is the window radius of the mean filter; represents the pixel value at the coordinate in the original underwater image under the L channel; is the enhanced image feature of the L channel; is the pixel value at the coordinate in the original underwater image under the L channel; n and m are the serial numbers of pixel points in the local pixel block; Send the original underwater image into the CIELab color space Channel sum Obtain the red-green chromaticity feature and yellow-cyan chromaticity feature from the channels; Overlay the L-channel enhanced image features, red-green chromaticity features, and yellow-cyan chromaticity features to obtain a locally enhanced image.

3. The underwater image enhancement method according to claim 1, wherein Use the Contrast Limited Adaptive Histogram Equalization method to perform color restoration on the original underwater image to obtain an equalized image, specifically including: Decompose the original underwater image into three-channel initial images; the three-channel initial images are the R-channel initial image, G-channel initial image, and B-channel initial image respectively; After performing grayscale processing on the channel initial images, divide the channel initial images into M×N sub-image blocks, and calculate the histograms of the sub-image blocks respectively. The expression formula is: ； ； ； In the formula, is the pixel at the coordinate after grayscale conversion of the initial channel image ; is the gray level, ; is the Kronecker function, and PS is the set of pixel point coordinates of the initial channel image; is the frequency of the pixel point with gray level k in the initial channel image; is the histogram of the sub-image block; c is the color channel; Perform a clipping operation on the histogram of each sub-image block according to the set clipping threshold to obtain a clipped histogram. The expression formula is: ； ； ； ； In the formula, is the shear threshold; is the shear coefficient; is the gray level of the sub-image block; is the total number of pixel values exceeding the shear threshold in the sub-image block; is the evenly distributed pixel value exceeding the shear threshold in the sub-image block; is the shear histogram; Perform bilinear interpolation reconstruction on the clipped histogram to obtain an equalized image.

4. The underwater image enhancement method according to claim 3, wherein Linearly weighted fuse the locally enhanced image and the equalized image to obtain a fused image, specifically including: Calculate the covariance matrix of the locally enhanced image. The expression formula is: ； ； Calculate the covariance matrix of the equalized image. The expression formula is: ； ； In the formula, is the covariance matrix of the locally enhanced image; is the covariance matrix of the equalized image; is the locally enhanced image; is the equalized image; is the average pixel value of the locally enhanced image; is the average pixel value of the equalized image; t is the index of the pixel point; Calculate the eigenvectors according to the covariance matrices of the locally enhanced image and the equalized image, and construct a principal component space from the eigenvectors of the locally enhanced image and the equalized image. The expression formula is: ； ； In the formula, is the covariance matrix 's eigenvalue; is the covariance matrix 's eigenvector; is the covariance matrix 's eigenvalue; is the covariance matrix 's eigenvector; is the principal component space corresponding to the locally enhanced image; is the principal component space corresponding to the equalized image; , , are the eigenvectors of the first, second, and third principal components in the locally enhanced image, respectively; , , are the eigenvectors of the first, second, and third principal components in the equalized image, respectively; Map the locally enhanced image and the equalized image to the principal component space to obtain key features. The expression formula is: ； In the formula, maps the locally enhanced image to the principal component space to obtain key features; maps the equalized image to the principal component space to obtain key features; is the locally enhanced image; is the equalized image; Calculate the fusion weight from the key features; perform image fusion on the locally enhanced image and the equalized image according to the fusion weight to obtain a fused image.

5. The underwater image enhancement method according to claim 4, wherein Calculate the fusion weight from the key features; perform image fusion on the locally enhanced image and the equalized image according to the fusion weight to obtain a fused image, specifically including: ； ； In the formula, is the fusion weight of the locally enhanced image; is the fusion weight of the equalized image; t is the index of the pixel point; is the principal component serial number; is the fused image.

6. The underwater image enhancement method according to claim 1, wherein, Perform color correction on the fused image in the CIELab color space channel and channel color correction to obtain a corrected image, specifically including: ； ； ； In the formula, is the corrected image feature of the fused image in channel, is the image feature of the fused image in channel, is the average value of the image features of the fused image in channel; is the corrected image feature of the fused image in channel, is the image feature of the fused image in channel, is the average value of the image features of the fused image in channel; is the corrected image; is the mapping function from CIELab color space to RGB color space; is the image feature of the fused image in channel, c is the channel number of RGB color space, R is the red channel of RGB color space, G is the green channel of RGB color space, and B is the green channel of RGB color space.

7. The underwater image enhancement method according to claim 5 or 6, characterized in that Perform image fusion on the corrected image, depth guidance map, and detail image to obtain the final enhanced image, specifically including: ； ； In the formula, is the final enhanced image; is the depth guidance map; is the corrected image; are the image features of image fusion in the three RGB channels; is the detail image; represents the Gaussian blur processing function.

8. An underwater image enhancement system, characterized in that, Including: An acquisition module for acquiring the original underwater image; A local enhancement module for performing L-channel contrast enhancement on the original underwater image in the CIELab color space to obtain a locally enhanced image; An image equalization module for performing color restoration on the original underwater image using the Contrast Limited Adaptive Histogram Equalization method to obtain an equalized image; A fusion module for linearly weighted fusing the locally enhanced image and the equalized image to obtain a fused image; A calibration module for performing channel and channel color correction on the fused image to obtain a corrected image; performing depth prediction on the fused image by a monocular depth estimation method to obtain a depth guidance map; performing RGB channel information expansion on the fused image to obtain a detail image; An output module for performing image fusion on the corrected image, depth guidance map, and detail image to obtain the final enhanced image.

9. An electronic terminal, characterized in that, It includes a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the underwater image enhancement method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it realizes the steps of the underwater image enhancement method according to any one of claims 1 to 7.

Images