Underwater image enhancement method and system based on binocular vision and polarization imaging

Abstract

The invention discloses an underwater image enhancement method and system based on binocular vision and polarization imaging. The method comprises the following steps: firstly, synchronously acquiring an orthogonal polarization image pair through a binocular polarization imaging unit, recovering scene depth information from the polarization image pair by using a binocular stereoscopic vision technology, and calculating a transmissivity graph; analyzing the polarization characteristic difference between target reflected light and backscattered light, establishing a polarization difference model, and separating background scattered light; solving an underwater imaging equation in combination with the depth information and a polarization difference model to obtain a preliminary restoration result of the target reflected light; and finally, further improving the image quality through the steps of multi-scale spectrum analysis, adaptive filter construction and post-processing by adopting a spectrum adaptive image enhancement technology. The system comprises a binocular polarization imaging unit, a calculation processing unit and an active illumination unit. According to the method, the problems of color distortion, low contrast and fuzzy details of the underwater image are solved, and the visual quality and usability of the underwater image are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer vision and digital image processing, and particularly relates to an underwater image enhancement method and system based on binocular vision and polarization imaging. BACKGROUND

[0002] When light propagates underwater, it will be absorbed and scattered by water and suspended particles in the water, resulting in serious quality degradation of underwater images. Absorption causes light intensity attenuation and color distortion, especially the rapid loss of red light energy in the long-wave band, causing the image to be severely blue-green. Scattering mainly includes forward scattering and back scattering: forward scattering causes light to deviate from the original path, resulting in image blurring and loss of details; back scattering forms a "curtain" of non-target reflected light between the camera and the target, which seriously reduces the image contrast and produces a fog-like effect.

[0003] Existing underwater image processing techniques mainly fall into two categories. The first category is based on physical model restoration, the core of which is to estimate water parameters and scene structure, such as depth or transmittance, and then reverse the image degradation model. However, in complex real environments, it is extremely challenging to accurately estimate these parameters, for example, statistical priors based on a single image, such as the dark channel prior, which can easily fail in areas with uneven lighting or white objects. The second category is based on image enhancement methods, such as histogram equalization, Retinex algorithm, wavelet transform, etc. This kind of method does not rely on physical model, mainly through adjusting the pixel distribution of the image or decomposing the light and reflection components to improve the visual effect, but they have obvious limitations: histogram equalization may over-enhance noise and cause color distortion; Retinex algorithm can produce halos at light-dark boundaries and amplify dark noise; if fixed filters are used in frequency domain methods such as wavelet transform, it is difficult to adapt to the changing underwater environment, and may introduce "pseudo-Gibbs" effect.

[0004] Polarization imaging technology provides a physical basis for separating signal light and scattered light by analyzing the differences in polarization characteristics between target reflected light and background scattered light. Binocular stereo vision technology can passively obtain accurate depth information of the scene. However, existing technologies have not effectively integrated the scattering light separation capability provided by polarization information, the accurate physical depth information provided by binocular vision, and the intelligent enhancement technology that can adapt to complex environments, forming a systematic and robust underwater image processing solution. SUMMARY

[0005] The application aims to provide an underwater image enhancement method and system based on binocular vision and polarization imaging, effectively fuse the scattering light separation ability provided by polarization information, the accurate physical depth information provided by binocular vision and the intelligent enhancement technology capable of adapting to complex environment, improve the visual quality of underwater images, and improve the recovery accuracy and image quality, and enhance the adaptability and generalization ability of underwater image processing method.

[0006] The technical solution for achieving the object of the application is an underwater image enhancement method based on binocular vision and polarization imaging, comprising the following steps:

[0007] Step 1, synchronously collecting two orthogonal polarization images of a target scene by a binocular imaging system with orthogonal polarizers, the two orthogonal polarization images being a first polarization image and a second polarization image ;

[0008] Step 2, calculating a disparity map based on the first polarization image and the second polarization image , and restoring a depth map of the scene according to a binocular vision imaging model ;

[0009] Step 3, calculating a transmittance map of the scene according to the depth map of the scene and the Beer-Lambert law ;

[0010] Step 4, performing fusion and difference processing on the first polarization image and the second polarization image to obtain a total light intensity map and a polarization difference map , and estimating background scattering light based on the transmittance map of the scene ;

[0011] Step 5, combining the total light intensity map , the transmittance map and the estimated background scattering light, and solving an underwater imaging physical model to restore a target reflected light image ;

[0012] Step 6, performing spectral adaptive image enhancement processing on the target reflected light image to obtain a final enhanced image.

[0013] Further, the two orthogonal polarization images of a target scene are synchronously collected by a binocular imaging system with orthogonal polarizers in step 1, the two orthogonal polarization images being a first polarization image and a second polarization image , and the specific steps are as follows:

[0014] An active illumination source is used, and a polarizer is placed in front of the source to make the light illuminating the target scene linearly polarized. In the binocular polarization imaging unit, a polarizer placed in front of one camera is parallel to the polarization direction of the illumination light, and is used to acquire the first polarized image. A polarizer positioned in front of another camera, perpendicular to the polarization direction of the illumination light, is used to acquire the second polarized image. .

[0015] Furthermore, step 2, based on the first polarization image Second polarization image The disparity map is calculated through stereo matching, and the depth map of the scene is reconstructed based on the binocular vision imaging model. The details are as follows:

[0016] With the first polarization image Based on the second polarization image Perform dense stereo matching to obtain a disparity map. The depth map is calculated using the following formula. :

[0017]

[0018] in, For camera focal length, The baseline distance of the binocular camera. The pixel size.

[0019] Furthermore, the transmittance map of the scene described in step 3. The calculation formula is:

[0020]

[0021] in, The attenuation coefficient of the water body.

[0022] Further, step 4 involves processing the first polarization image. Second polarization image The total intensity map is obtained by performing fusion and differential processing. and polarization difference diagram And based on the scene's transmittance map The background scattered light is estimated as follows:

[0023] Calculate the total light intensity map ;

[0024] Calculate the polarization difference plot ;

[0025] Background scattered light The estimated satisfy the following relation: wherein is the background light intensity at infinity.

[0026] Further, the target reflected light image in step 5 is The calculation formula is as follows:

[0027]

[0028] wherein, is the total light intensity map.

[0029] Further, the target reflected light image in step 6 is processed by the spectral adaptive image enhancement processing to obtain the final enhanced image, and the specific process is as follows:

[0030] Step 6.1, spectral parameter analysis: input a pair of registered background image and target image under the same water area, perform multi-scale spectral analysis through Fourier transform and wavelet transform, quantify the frequency domain feature difference between the background and the target, calculate the adaptive filter parameters, and automatically select the optimal filtering strategy based on the preset rules;

[0031] Step 6.2, intelligent filtering processing: Fourier transform is performed on the input image to be enhanced, an adaptive frequency domain filter is constructed according to the parameters obtained in step 6.1 and the selected strategy, and the image is reconstructed through inverse Fourier transform after frequency domain filtering;

[0032] Step 6.3, post-processing: the filtered image is sequentially subjected to gamma correction and limited contrast adaptive histogram equalization processing to obtain the final enhanced image.

[0033] Further, the filtering strategy in step 6.1 includes at least one of the following:

[0034] Difference enhancement strategy, focusing on amplifying the amplitude difference between the target and the background in all frequency bands;

[0035] Background suppression strategy, focusing on directly suppressing the low-energy spectral region identified as the background;

[0036] Target extraction strategy, focusing on enhancement in specific medium-high frequency bands to highlight the target edge and texture.

[0037] An underwater image enhancement system based on binocular vision and polarization imaging, which is used to realize the underwater image enhancement method based on binocular vision and polarization imaging, and the system comprises a binocular polarization imaging unit, a calculation processing unit and an active illumination unit.

[0038] The binocular polarization imaging unit comprises a first synchronous camera and a second synchronous camera, and the front of each of the cameras is respectively provided with a first polarizer and a second polarizer with mutually orthogonal light transmission axis directions, and is used for synchronously collecting a pair of orthogonal polarization images;

[0039] The computing processing unit is used for receiving the pair of orthogonal polarization images, performing stereo matching and depth recovery, calculating a transmittance map, performing polarization analysis and scattered light estimation, restoring a target reflected light image, executing a spectral adaptive image enhancement algorithm, and outputting a final enhanced image.

[0040] The active illumination unit comprises a light source and a third polarizer arranged in front of the light source, and is used for generating linearly polarized light to irradiate a target scene.

[0041] Further, the direction of the first polarizer of the first synchronous camera in the binocular polarization imaging unit is parallel to the polarization direction of the third polarizer of the active illumination unit, and the direction of the second polarizer of the second synchronous camera is perpendicular to the polarization direction of the third polarizer of the active illumination unit.

[0042] Compared with the prior art, the present application has the following advantages:

[0043] (1) unification of physical accuracy and adaptability: the present application combines the polarization physical model-based restoration method and the data-driven spectral adaptive enhancement technology, the former guarantees the physical authenticity and reliability of the restoration process, and the latter endows strong adaptability to complex and variable underwater environments, overcoming the poor generalization ability of the traditional fixed parameter method;

[0044] (2) accuracy of depth information utilization: the physical depth of the scene is directly obtained through binocular stereo vision, and then a more actual transmittance map is calculated, which has higher accuracy compared with the depth estimation method based on image color or texture statistics prior, especially in the area with sudden scene depth change or large-area white target, which can effectively avoid restoration distortion;

[0045] (3) efficient scattered light separation: the essential difference between target reflected light and background scattered light in polarization characteristics is fully utilized, and through orthogonal polarization imaging and differential operation, the backscattered light can be effectively separated and suppressed, thereby improving the image signal-to-noise ratio from the source;

[0046] (4) intelligent and adaptive enhancement strategy: the spectral adaptive enhancement method is adopted, the optimal filter is intelligently selected and constructed by analyzing the frequency domain feature "fingerprint" of the background and the target, which can enhance the target details and textures while maximally suppressing the background noise, thereby avoiding the problems such as halo, artifacts and noise amplification often introduced by the traditional enhancement method;

[0047] (5) Multi-function information output: this method not only outputs high-quality restored and enhanced images, but also synchronously generates accurate depth map, transmittance map and polarization degree map, which provides valuable data support for underwater robot navigation, target three-dimensional reconstruction, material identification and other advanced visual tasks. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 is a flowchart of a kind of underwater image enhancement method based on binocular vision and polarization imaging of the present application.

[0049] Figure 2 is a structural schematic diagram of a kind of underwater image enhancement system based on binocular vision and polarization imaging of the present application.

[0050] Figure 3 is the image effect comparison chart after different method processing in the embodiment of the present application, from left to right, it is original drawing, Retinex algorithm, Curvelet transformation, the image processed by the method of the present application. DETAILED DESCRIPTION

[0051] The present application will be further described in detail below in combination with the drawings and specific embodiments.

[0052] As Figure 1 shown, a kind of underwater image enhancement method based on binocular vision and polarization imaging of the present application includes the following steps:

[0053] Step 1, through the binocular imaging system with orthogonal polarizer, the two orthogonal polarization images of target scene are synchronously collected, respectively first polarization image And second polarization image ;

[0054] Step 2, based on first polarization image And second polarization image , calculate disparity map through stereo matching, and restore the depth map of scene according to binocular vision imaging model ;

[0055] Step 3, according to the depth map of scene And Beer-Lambert law, calculate the transmittance map of scene ;

[0056] Step 4, first polarization image And second polarization image Fusion and difference processing are carried out, and total light intensity map And polarization difference map , and estimate background scattered light based on the transmittance map of scene ;

[0057] Step 5: Combine with the total light intensity diagram Transmittance diagram By solving the underwater imaging physical model and estimating the background scattered light, the target reflected light image is reconstructed. ;

[0058] Step 6: Image of the reflected light from the target Perform spectral adaptive image enhancement processing to obtain the final enhanced image.

[0059] As a specific example, step 1 involves using a binocular imaging system with orthogonal polarizers to simultaneously acquire two orthogonally polarized images of the target scene, namely the first polarization image. Second polarization image The details are as follows:

[0060] An active illumination source is used, and a polarizer is placed in front of the source to make the light illuminating the target scene linearly polarized. In the binocular polarization imaging unit, a polarizer placed in front of one camera is parallel to the polarization direction of the illumination light, and is used to acquire the first polarized image. A polarizer positioned in front of another camera, perpendicular to the polarization direction of the illumination light, is used to acquire the second polarized image. .

[0061] As a specific example, step 2, based on the first polarization image Second polarization image The disparity map is calculated through stereo matching, and the depth map of the scene is reconstructed based on the binocular vision imaging model. The details are as follows:

[0062] With the first polarization image Based on the second polarization image Perform dense stereo matching to obtain a disparity map. Based on the binocular vision imaging model, and combining camera intrinsic and extrinsic parameters, the depth map is calculated using the following formula. :

[0063]

[0064] in, For camera focal length, The baseline distance of the binocular camera. The pixel size.

[0065] As a specific example, step 3 involves using the depth map of the scene. Using Beer-Lambert's law, calculate the transmittance map of the scene. The calculation formula is:

[0066]

[0067] in, The attenuation coefficient of the water body can be set according to the turbidity of the water body or obtained through auxiliary measurement.

[0068] As a specific example, step 4 describes the first polarization image Second polarization image The total intensity map is obtained by performing fusion and differential processing. and polarization difference diagram And based on the scene's transmittance map The background scattered light is estimated as follows:

[0069] Calculate the total light intensity map ;

[0070] Calculate the polarization difference plot ;

[0071] Background scattered light The estimated satisfy the following relation: ,in The background light intensity at infinity can usually be estimated from the brightest region in the image; the polarization difference ΔB of the backscattered light and its total intensity B satisfy the following:

[0072]

[0073] in, The polarization degree of the backscattered light is a parameter related to the characteristics of water particles.

[0074] As a specific example, step 5 involves combining the total light intensity map. Transmittance diagram By solving the underwater imaging physical model and estimating the background scattered light, the target reflected light image is reconstructed. The details are as follows:

[0075] The underwater imaging model is as follows:

[0076]

[0077] The result obtained in step 3 and the result obtained in step 4 By substituting the values ​​and utilizing the constraints provided by polarization difference, the true reflected light from the target can be calculated. The formula is:

[0078]

[0079] in, This is the total light intensity diagram.

[0080] As a specific example, step 6 describes the image of the reflected light from the target. Spectrum-adaptive image enhancement processing is performed to obtain the final enhanced image, as follows:

[0081] Step 6.1, Spectrum Parameter Analysis: Input a pair of registered background and target images in the same water area, perform multi-scale spectrum analysis through Fourier transform and wavelet transform, quantify the frequency domain feature differences between the background and the target, calculate the adaptive filter parameters, and automatically select the optimal filtering strategy based on preset rules;

[0082] The filtering strategy includes at least one of the following:

[0083] The difference enhancement strategy focuses on comprehensively amplifying the amplitude differences between the target and the background across all frequency bands;

[0084] Background suppression strategies focus on directly suppressing low-energy spectral regions identified as background.

[0085] The target extraction strategy focuses on enhancing specific mid-to-high frequency bands to highlight target edges and textures.

[0086] Step 6.2, Intelligent Filtering Processing: Perform Fourier transform on the input image to be enhanced, construct an adaptive frequency domain filter based on the parameters obtained in Step 6.1 and the selected strategy, and reconstruct the image through inverse Fourier transform after frequency domain filtering;

[0087] Step 6.3, Post-processing: Perform gamma correction and contrast-limited adaptive histogram equalization on the filtered image in sequence to obtain the final enhanced image.

[0088] like Figure 2 As shown, the present invention also provides an underwater image enhancement system based on binocular vision and polarization imaging, including a binocular polarization imaging unit, a computing processing unit, and an active illumination unit;

[0089] The binocular polarization imaging unit includes a first synchronous camera 6 and a second synchronous camera 7, each equipped with a first polarizer 4 and a second polarizer 5 whose transmission axes are orthogonal to each other, for synchronously acquiring orthogonal polarization image pairs.

[0090] The computational processing unit is used to receive the orthogonal polarization image pairs, perform stereo matching and depth restoration, calculate the transmittance map, perform polarization analysis and scattered light estimation, restore the target reflected light image, execute the spectrum adaptive image enhancement algorithm, and output the final enhanced image.

[0091] The active illumination unit includes a light source 1 and a third polarizer 2 placed in front of the light source, used to generate linearly polarized light to illuminate the target scene.

[0092] As a specific example, in the binocular polarization imaging unit, the direction of the first polarizer 4 of the first synchronous camera 6 is parallel to the polarization direction of the third polarizer 2 of the active illumination unit, and the direction of the second polarizer 5 of the second synchronous camera 7 is perpendicular to the polarization direction of the third polarizer 2 of the active illumination unit.

[0093] Example

[0094] The invention was implemented in an experimental water tank, where a milky white suspension was added to simulate moderate turbidity conditions.

[0095] 1. System Setup and Calibration

[0096] Hardware: Active illumination uses a high-brightness white light source 1, with a third polarizer 2 added in front. The binocular imaging system uses two industrial CCD cameras of the same model and parameters, mounted on a rigid bracket, with the baseline distance precisely set to 12cm. A first polarizer 4 is installed in front of the first synchronous camera 6, and a second polarizer 5 is installed in front of the second synchronous camera 7. The computing unit is a laptop computer.

[0097] Software and Calibration: Camera calibration was implemented using the OpenCV library. The Zhang Zhengyou calibration method was employed, and approximately 15 chessboard images with different poses were captured to obtain the intrinsic parameter matrices, distortion coefficients, and rotation matrix between the left and right cameras. Translation vector Subsequently, based on the Bouquet algorithm, the stereoRectify function is called to calculate the stereo correction parameters and generate a remapping table to ensure that the subsequently acquired left and right images are aligned in coplanar rows.

[0098] 2. Image Acquisition

[0099] A set of PVC simulated underwater biological models was placed in the center of the pool. The LED active light source was turned on, simultaneously triggering a binocular camera to acquire orthogonal polarization image pairs. and . It is relatively bright, and the scattered light has a significant impact. It's darker, so the target information is purer.

[0100] 3. Image Restoration Process

[0101] Depth and transmittance: based on corrected values The image on the left is shown. As shown in the right image, the disparity map is obtained by performing dense stereo matching using the Semi-Global Matching algorithm. According to camera parameters =25mm, =0.12m, =4.5× m and the formula are used to calculate the depth map. Set the water attenuation coefficient β=0.2, and calculate the transmittance map according to the Beer-Lambert law. .

[0102] Polarization restoration: Calculation of total intensity map ;from The brightest area in an image is usually the background at infinity, which is estimated to be... Based on experimental environment experience, the settings were configured. =0.8, substitute all the above parameters into the restoration formula The preliminary restored image was obtained.

[0103] 4. Image Enhancement Process

[0104] Offline spectral parameter analysis was performed: An image containing only background water and another image containing a clear PVC crab model were pre-acquired as analysis pairs. Fourier transforms were performed on both images, the logarithmic amplitude spectrum was calculated, and frequency bands were divided. Analysis revealed that the target's intensity differed from the background in the mid-to-high frequency band (0.25 ≤ r < 0.4). The algorithm automatically selected "target extraction" as the optimal filtering strategy and calculated the corresponding enhancement factor, as the target was the largest. =1.8 and inhibitory factor =0.3.

[0105] Intelligent filtering and post-processing: The image restored in step 3 The pre-constructed target extraction filter is applied, multiplied by 1.8 within the frequency band [0.25, 0.4], and multiplied by 0.3 in other frequency bands. An inverse Fourier transform is then performed after filtering. Subsequently, gamma correction with γ=0.75 is applied to moderately brighten the image, followed by CLAHE processing with a grid size of 8x8 and a limiting coefficient of 2.0 to obtain the final enhanced image.

[0106] 5. Results and Analysis

[0107] The final result of this embodiment is compared with the original image and the result processed using the traditional Retinex algorithm and Curvelet transform.

[0108] Subjective evaluation: such as Figure 3 As shown, the method of the present invention performs best in terms of edge sharpness, detail preservation, and color naturalness, and the background is clean with no obvious halo or noise.

[0109] Objective evaluation: Quantitative evaluation was conducted using the no-reference image quality metrics UIQM and UCIQE. The results are as follows: Figure 3 As shown in Table 1:

[0110] Table 1

[0111] Image processing method UIQM UCIQE Original image 9.9165 0.4655 Retinex algorithm 7.6239 0.5256 Curvelet transform 7.7120 0.5030 Method of the invention 10.5867 0.5484

[0112] As shown in Table 1, the method of this invention outperforms the comparative methods in both metrics, achieving the highest overall image quality, fully demonstrating the effectiveness and superiority of this invention. This invention not only significantly improves the visual quality of underwater images but also provides a reliable data foundation for subsequent visual applications.

[0113] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An underwater image enhancement method based on binocular vision and polarization imaging, characterized in that, The method comprises the following steps: Step 1, synchronously collecting two orthogonal polarization images of a target scene by a binocular imaging system with orthogonal polarizers, which are respectively a first polarization image and a second polarization image ; Step 2, calculating a disparity map based on the first polarized image and the second polarized image , and recovering a depth map of the scene according to a binocular vision imaging model ; Step 3, Compute the transmittance map of the scene from its depth map and the Beer-Lambert law ; Step 4, fusion of the first polarized image and the second polarized image with difference processing to obtain total intensity image and polarized difference image and based on the scene transmittance map estimate the background scattered light; Step 5, combining total light intensity map , transmittance map and estimated background scattered light, the target reflectance image is recovered by solving the underwater imaging physics model ; Step 6, performing spectral adaptive image enhancement processing on the target reflected light image spectrum adaptive image enhancement processing to obtain a final enhanced image.

2. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 1, characterized in that, The two orthogonal polarization images of the target scene are synchronously collected by the binocular imaging system with orthogonal polarizers according to step 1, and the first polarization image and the second polarization image are respectively Specifically as follows:​ The active illumination light source is adopted, and a polarizer is arranged in front of the light source, so that the light irradiated to the target scene is linearly polarized light; in the binocular polarization imaging unit, the polarizer arranged in front of one camera is parallel to the polarization direction of the illumination light, and is used for collecting the first polarization image ; and the polarizer arranged in front of the other camera is perpendicular to the polarization direction of the illumination light, and is used for collecting the second polarization image .

3. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 2, characterized in that, the first polarization image and the second polarization image , the parallax map is calculated by stereo matching and the depth map of the scene is recovered according to a binocular vision imaging model , specifically as follows: with the first polarized image as a reference, performing dense stereo matching on the second polarized image to obtain a disparity map and calculating a depth map by the following formula : ； wherein, is the camera focal length, is the baseline distance of the binocular camera, is the pixel size.

4. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 3, characterized in that, transmittance map of the scene described in step 3 The formula for calculating is: ； wherein, is the attenuation coefficient of the water body.

5. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 4, characterized in that, the first polarized image described in step 4 and the second polarized image perform fusion and difference processing to obtain total light intensity image and polarization difference image and the transmittance map of the scene estimate the background scattered light, in particular as follows: calculating a total light intensity map ; Computing polarization difference maps ; Background scattered light The estimate of the background scattered light satisfies the relation: where is the background light intensity at infinity.

6. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 5, characterized in that, the target reflected light image described in step 5 The formula for calculating is: ； wherein is the total light intensity map.

7. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 6, characterized in that, the target reflected light image described in step 6 The spectrum adaptive image enhancement processing is performed to obtain the final enhanced image, and the specific process is as follows: Step 6.1, spectral parameter analysis: input a pair of background image and target image registered under the same water area, perform multi-scale spectral analysis through Fourier transform and wavelet transform, quantify the frequency domain feature difference between the background and the target, calculate adaptive filter parameters, and automatically select the optimal filtering strategy based on preset rules; Step 6.2, intelligent filtering processing: perform Fourier transform on the input image to be enhanced, construct an adaptive frequency domain filter according to the parameters obtained in step 6.1 and the selected strategy, perform frequency domain filtering, and reconstruct the image through inverse Fourier transform; Step 6.3, post-processing: perform gamma correction and limited contrast adaptive histogram equalization on the filtered image in sequence to obtain the final enhanced image.

8. The underwater image enhancement method based on binocular vision and polarization imaging according to claim 7, characterized in that, The filtering strategy in step 6.1 includes at least one of the following: Difference enhancement strategy, focusing on amplifying the amplitude difference between the target and the background in all frequency bands; Background suppression strategy, focusing on directly suppressing the low-energy spectral region identified as background; Target extraction strategy, focusing on enhancing in specific medium-high frequency bands to highlight target edges and textures.

9. An underwater image enhancement system based on binocular vision and polarization imaging, characterized in that, The system is used to implement the underwater image enhancement method based on binocular vision and polarization imaging according to any one of claims 1-8, and the system comprises a binocular polarization imaging unit, a computing processing unit and an active illumination unit; The binocular polarization imaging unit comprises a first synchronous camera (6) and a second synchronous camera (7), and the front of each camera is provided with a first polarizer (4) and a second polarizer (5) with mutually orthogonal light transmission axis directions, respectively, for synchronously collecting a pair of orthogonal polarization images; The computing processing unit is used to receive the pair of orthogonal polarization images, perform stereo matching and depth recovery, calculate a transmittance map, perform polarization analysis and scattered light estimation, restore a target reflected light image, execute a spectral adaptive image enhancement algorithm, and output a final enhanced image; The active illumination unit comprises a light source (1) and a third polarizer (2) placed in front of the light source, and is used to generate linearly polarized light to illuminate a target scene.

10. The binocular vision and polarization imaging based underwater image enhancement system according to claim 9, wherein, The direction of the first polarizer (4) of the first synchronous camera (6) in the binocular polarization imaging unit is parallel to the polarization direction of the third polarizer (2) of the active illumination unit, and the direction of the second polarizer (5) of the second synchronous camera (7) is perpendicular to the polarization direction of the third polarizer (2) of the active illumination unit.

Images