An underwater image enhancement method and system based on polarization difference imaging

Abstract

The present invention provides an underwater image enhancement method and system based on polarization difference imaging, which relates to the technical field of image processing. The method includes obtaining a depth image corresponding to an underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and calculating a difference image between the two polarization images; based on the difference image, respectively dividing the two polarization images into regions, and calculating the background light intensity of each divided region by using the depth image; based on the background light intensity, calculating and optimizing the perspective rate of each pixel point in the corresponding region; restoring the image content of the corresponding region of the corresponding polarization image based on the perspective rate, and performing weighted fusion and global image enhancement processing on the two restored polarization images according to the polarization degree to obtain a final enhanced image. By combining the information of the depth image and the difference image, the present invention can more accurately estimate the background light intensity and perspective rate of each region, thereby realizing effective enhancement of underwater images.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly to an underwater image enhancement method and system based on polarization differential imaging. Background Art

[0002] In the field of underwater imaging, due to the absorption, scattering of water and the change of illumination conditions, it has always been a technical problem to obtain clear and high-quality underwater images. Traditional underwater image enhancement methods, such as histogram equalization, contrast stretching, etc., although they can improve the visual effect of the image to a certain extent, often cannot effectively remove the blurring and color distortion caused by the underwater environment. In recent years, with the development of polarization imaging technology, the method of using polarization information to enhance underwater images has gradually become a research hotspot.

[0003] Polarization imaging technology can provide more information than traditional imaging by capturing the polarization state of light waves, especially having significant advantages in dealing with scattered light and background light interference. In the underwater environment, due to the existence of scattered light, it is often difficult to distinguish the target and background in the image. And polarization imaging technology can distinguish scattered light and direct light by measuring the polarization state of light waves, thereby effectively reducing the influence of scattered light and improving the clarity of the image.

[0004] However, most of the existing underwater image enhancement methods based on polarization imaging only utilize the image information of a single polarization state, ignoring the differences between images in different polarization states, which limits the improvement of the image enhancement effect. In addition, when dealing with complex underwater scenes, these methods often have difficulty in accurately estimating the background light intensity and transmittance, resulting in deviations in the image restoration results.

[0005] Therefore, it is necessary to provide an underwater image enhancement method and system based on polarization differential imaging to solve the above technical problems. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides an underwater image enhancement method and system based on polarization differential imaging. By obtaining a depth image corresponding to the underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and using the difference image between the two polarization images for region division and calculation of the background light intensity. By combining the information of the depth image and the difference image, this method can more accurately estimate the background light intensity and transmittance of each partition, thereby realizing effective enhancement of the underwater image.

[0007] The present invention provides an underwater image enhancement method based on polarization differential imaging, and the method includes the following steps:

[0008] Obtain a depth image corresponding to the underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and calculate a difference image between the two polarization images;

[0009] Based on the difference image, divide the two polarization images into regions respectively, and calculate the background light intensity of each divided region by using the depth image;

[0010] Based on the background light intensity, calculate and optimize the perspective rate of each pixel point in the corresponding region;

[0011] Based on the perspective rate, restore the image content of the corresponding region of the corresponding polarization image, and perform weighted fusion and global image enhancement processing on the two restored polarization images according to the polarization degree to obtain the final enhanced image.

[0012] Preferably, the divided regions include a target region, a background region, and a transition region.

[0013] Preferably, the acquisition of the polarization image includes:

[0014] Use the same shooting device to sequentially obtain two polarization images at different polarization degrees, and the polarization degree difference exceeds the set threshold.

[0015] Preferably, the acquisition of the difference image includes:

[0016] For the two polarization images, perform pixel-level matching at the same spatial position;

[0017] For each pair of matched pixel points, calculate the difference between the pixel values of the two polarization images, and take the absolute value to obtain the difference image.

[0018] Preferably, the step of dividing the two polarization images into regions respectively based on the difference image includes:

[0019] Process the difference image based on a semantic segmentation model of deep learning to identify and mark the target region, the background region, and the transition region;

[0020] Map the marking result to the corresponding two polarization images, and divide the two polarization images into a target region, a background region, and a transition region.

[0021] Preferably, the step of calculating the background light intensity of each divided region by using the depth image includes:

[0022] For the background region, extract the pixel difference value of the corresponding region in the difference image, calculate the variance of the pixel difference value and use it as a weight factor, and perform weighted averaging on the pixel values of the depth image to obtain the background light intensity of the background region;

[0023] For the target region, the polarization attenuation coefficient is determined according to the ratio of the degrees of polarization of the two polarization images in the target region, and combined with the local gradient information of the depth image, an adaptive weight is generated through a normalized product function to calculate the background light intensity of the target region;

[0024] For the transition region, edge detection is performed on the difference image and compared with a threshold of the pixel intensity difference between the two polarization images to generate a mask, and based on the mask, bilinear interpolation is performed on the background light intensities of the background region and the target region to obtain the background light intensity of the transition region.

[0025] Preferably, calculating and optimizing the transmittance of each pixel point in the corresponding partition based on the background light intensity includes:

[0026] For each pixel point in each partition, a preliminary transmittance is calculated based on the background light intensity and the original brightness value of the pixel point in the polarization image, and the calculation formula is:

[0027]

[0028] where, represents the background light intensity of the pixel point and the value range is , represents the original brightness value of the pixel point , represents the preliminary transmittance of the pixel point ;

[0029] The preliminary transmittance is optimized according to the local gradient value of the depth image and the pixel difference value of the difference image to obtain the final transmittance, and the optimization formula is:

[0030]

[0031] where, and are respectively preset weight parameters for controlling the influence of local gradient and pixel difference, and respectively represent the maximum values of the local gradient of the depth image and the pixel difference value of the difference image, and respectively represent the local gradient value and the pixel difference value of the pixel point .

[0032] Preferably, the restoration of the image content includes:

[0033] For each pixel point in each partition, the brightness value of the corresponding pixel point in the polarization image is restored through an image restoration formula using the optimized transmittance and the background light intensity.

[0034] Preferably, the weighted fusion of the two restored polarization images according to the degree of polarization includes:

[0035] Obtain the degree-of-polarization difference value of each pixel point in the two polarization images, and query the corresponding weight factor according to a predefined mapping relation table, where the predefined mapping relation table divides the degree-of-polarization difference value into multiple continuous intervals, and each interval corresponds to a preset weight factor;

[0036] Based on the queried weight factor, perform weighted superposition on the brightness values of the corresponding pixel points in the two restored polarization images to generate a fused image.

[0037] The present invention also provides an underwater image enhancement system based on polarization difference imaging for performing an underwater image enhancement method based on polarization difference imaging. The system includes:

[0038] An image acquisition module for acquiring a depth image corresponding to the underwater image to be enhanced and two polarization images with a degree of polarization exceeding a set threshold, and calculating a difference image between the two polarization images;

[0039] A background light intensity calculation module for respectively dividing the two polarization images into regions based on the difference image, and calculating the background light intensity of each divided region by using the depth image;

[0040] A transmittance calculation module for calculating and optimizing the transmittance of each pixel point in the corresponding region based on the background light intensity;

[0041] An image enhancement module for restoring the image content of the corresponding region of the corresponding polarization image based on the transmittance, and performing weighted fusion and global image enhancement processing on the two restored polarization images according to the degree of polarization to obtain a final enhanced image.

[0042] Compared with the related art, an underwater image enhancement method and system based on polarization difference imaging provided by the present invention have the following beneficial effects:

[0043] The present invention first performs semantic segmentation on the difference image by using a deep learning model to divide the image into a target region, a background region, and a transition region. Then, according to the information of the depth image and the difference image, the background light intensity of each region is calculated respectively. Next, based on the background light intensity and the original brightness value of the pixel point, the transmittance of each pixel point is calculated and optimized. Finally, the brightness value of the polarization image is restored by using the optimized transmittance and the background light intensity, and weighted fusion and global image enhancement processing are performed on the two restored polarization images according to the degree of polarization to obtain a final enhanced image.

[0044] The present invention not only fully utilizes the advantages of polarization imaging technology, but also realizes more accurate enhancement of underwater images by combining the information of depth images and difference images. Description of the Drawings

[0045] Figure 1 It is a flowchart of an underwater image enhancement method based on polarization differential imaging provided by the present invention;

[0046] Figure 2 It is a module structure diagram of an underwater image enhancement method based on polarization differential imaging provided by the present invention. Detailed Embodiments

[0047] The present invention will be further described in detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. Additionally, it should be noted that for the convenience of description, only parts related to the present invention are shown in the drawings rather than all the structures. Furthermore, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0048] It should also be noted that for the convenience of description, only parts related to the present invention are shown in the drawings rather than all the content. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. When the operations are completed, the process can be terminated, but there may also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0049] Embodiment 1

[0050] The present invention provides an underwater image enhancement method based on polarization differential imaging. Referring to Figure 1 as shown, the method includes the following steps:

[0051] S1: Obtain a depth image corresponding to the underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and calculate the difference image between the two polarization images.

[0052] In this embodiment, through active imaging technologies including but not limited to structured light projection, binocular stereo vision, or ToF (Time-of-Flight) cameras, a depth image corresponding to the underwater image to be enhanced is synchronously acquired. Each pixel value in the depth image represents the distance from the corresponding point in the scene to the camera. It is necessary to ensure that the spatial resolution of the depth image is consistent with that of the polarization image, and the coordinate systems are aligned through calibration parameters.

[0053] Use an underwater camera with an adjustable polarizer, fix the shooting angle and lighting conditions, rotate the polarizer to two orthogonal directions (0° and 90°), and take two polarization images respectively. The polarization degree difference should be greater than or equal to 30%.

[0054] Perform image registration on the two polarization images to eliminate the displacement error caused by the shooting angle or movement. Use the SIFT feature point matching method to align the two images to ensure that each pixel corresponds to the same scene position in the two images.

[0055] For the two registered polarization images, calculate the absolute value of the intensity difference pixel by pixel to generate a difference image. Its pixel value reflects the local scattering difference between the two polarization images. The high-value area corresponds to the target edge or the dense area of suspended particles.

[0056] S2: Based on the difference image, divide the two polarization images into regions respectively, and calculate the background light intensity of each divided region using the depth image.

[0057] In step S2, the region division specifically includes:

[0058] First, process the difference image using a semantic segmentation model based on deep learning to identify and mark the target region, background region, and transition region.

[0059] In this embodiment, for the construction and training of the semantic segmentation model, it is necessary to collect an underwater scene polarization difference image dataset containing different turbidity, lighting conditions, and target types, and manually annotate the target region (difference value ≥ 60 and matching the depth contour), background region (difference value ≤ 30 and depth variance < 5%), and transition region (continuous change in difference and related to depth gradient). At the same time, rotate the data, simulate light attenuation, and enhance Gaussian noise to improve generalization. The model uses an improved U-Net architecture. The encoder uses ResNet-18 to extract features, and the decoder gradually restores the resolution through transposed convolution and outputs a three-channel probability map. The activation function uses ReLU in the encoder and LeakyReLU (negative slope 0.2) in the decoder to balance gradient stability. During the training process, use weighted cross-entropy loss (target weight 1.5, background 1.0, transition 2.0) and edge-sensitive regularization terms for joint optimization, and train through the AdamW optimizer with cosine annealing scheduling. The early stopping strategy prevents overfitting.

[0060] During inference, perform threshold segmentation on the output probability map (target region > 0.7, background region > 0.6) and filter out isolated regions with an area less than 50 pixels, and then refine the edges of the transition region through dilation and erosion operations (kernel size 5×5).

[0061] Next, map the marked results to the corresponding two polarization images, and divide the two polarization images into target regions, background regions, and transition regions.

[0062] In this embodiment, through the pre-calibrated camera internal parameters (focal length, principal point) and external parameters (polarizer rotation angle), establish the pixel-level coordinate mapping relationship between the difference image and the two polarization images. Then, map the segmentation results (target, background, transition region masks) of the difference image to the coordinate systems of the two polarization images through bilinear interpolation. During the mapping process, it is necessary to calculate the average polarization degree difference in the target region of the two polarization images. If the deviation from the average difference in the target region of the difference image exceeds 15%, it is determined that the mapping is incorrect, and re-registration is triggered. Verify the depth variance of the background region (required to be <5%); otherwise, adjust the segmentation threshold. Finally, output a three-region mask with the same resolution as the polarization image, where the target region is used for detail restoration, the background region suppresses scattering, and the transition region realizes smooth connection.

[0063] In step S2, the calculation of the background light intensity for each partition is specifically as follows:

[0064] For the background region, extract all pixel difference values belonging to the background region from the difference image, and calculate the variance of these difference values. The variance reflects the scattering uniformity of the background region: the smaller the variance, the more concentrated the difference distribution in the background region (i.e., the more uniform the water body scattering), and at this time, a higher weight is given.

[0065] When calculating the weight, take the reciprocal of the variance as the weight factor (if the variance is close to zero, add a minimum value to prevent division-by-zero errors). The larger the weight, the higher the contribution of the corresponding depth value to the background light intensity.

[0066] Weight-average the pixel values in the background region of the depth image according to the above weights to finally obtain the background light intensity.

[0067] For the target region, calculate the average intensity ratio of the target region in the two polarization images to obtain the polarization degree ratio. The range of this ratio is between 0 and 1, which is used to quantify the modulation ability of the target region to polarized light. The higher the ratio, the more significant the influence of the reflection or scattering on the target surface on the polarization characteristics.

[0068] And based on the polarization degree ratio, derive the polarization attenuation coefficient, which is used to quantify the attenuation degree of the reflection on the target surface to the polarization characteristics. The value range is limited to 0.2 - 0.8 (to avoid physical contradictions).

[0069] Among them, the calculation of the polarization attenuation coefficient incorporates the experimentally calibrated environmental polarization interference factor, which is used to correct external interferences such as water turbidity and lighting conditions. In a controllable experimental environment (such as a clear water tank and a turbid water tank), by measuring the known reflection characteristics of a standard target (such as a gray scale plate), comparing the theoretical polarization degree ratio with the actual imaging results, the environmental polarization interference factor is obtained by fitting. Finally, the polarization degree ratio is multiplied by the environmental interference factor to obtain the polarization attenuation coefficient.

[0070] The Sobel operator is used to calculate the local gradient of the depth image. The larger the gradient value, the more significant the geometric change of the target surface. Multiply the polarization attenuation coefficient by the normalized depth gradient to generate an adaptive weight. This weight assigns higher priority to the geometrically complex regions (high gradients) of the target surface, ensuring that the calculation of the background light intensity is more in line with the actual physical structure. The pixel values in the target region of the depth image are weighted and averaged according to the weight to obtain the background light intensity of the target region.

[0071] For the transition region, perform Canny edge detection on the difference image (Gaussian kernel = 1.5, low / high thresholds are 50 and 150 respectively) to mark the boundaries with significant scattering changes. Calculate the pixel intensity difference between the two polarization images, and retain the regions where the difference is greater than 30% of the maximum difference. Take the intersection with the edge detection result to generate a candidate mask for the transition region. Perform a closing operation (kernel size 3×3) on the mask to fill the holes, and take the intersection with the original segmentation result to eliminate noise interference.

[0072] Next, for each pixel in the transition region, calculate the Euclidean distances to the nearest background region and target region respectively. The closer the distance, the higher the weight of the background light intensity of the corresponding region. According to the distance weights, linearly interpolate the background light intensities of the background and target regions to obtain the light intensity value of the transition region, ensuring a smooth transition between different regions.

[0073] S3: Calculate and optimize the transmittance of each pixel point in the corresponding partition based on the background light intensity.

[0074] In this embodiment, for each pixel point in each partition, calculate the preliminary transmittance based on the background light intensity and the original brightness value of the pixel point in the polarization image. The calculation formula is as follows:

[0075]

[0076] Among them, represents the original brightness value of the pixel point , represents the preliminary transmittance of the pixel point , that is, the proportion of the light that is not scattered and absorbed by the water body during the process from the object surface to the camera, represents the background light intensity of the pixel point , and the value range is , represents the background light intensity of the pixel point, that is, the environmental light influence caused by scattering and absorption.

[0077] In this embodiment, the numerator in this formula represents the net brightness value after removing the background light; the denominator ensures that when the background light approaches 1 (i.e., the environmental light is very strong), the transmittance tends to zero, which conforms to the physical reality. This formula essentially estimates the actual transmitted light intensity by subtracting the background light.

[0078] Optimize the preliminary transmittance according to the local gradient value of the depth image and the pixel difference value of the difference image to obtain the final transmittance, where the optimization formula is:

[0079]

[0080] where, and are respectively preset weight parameters for controlling the influence of local gradient and pixel difference, , and respectively represent the maximum values of the local gradient of the depth image and the pixel difference of the difference image, and respectively represent the local gradient value and the pixel difference value of the pixel point .

[0081] In this embodiment, places with larger local gradient values usually correspond to target boundaries or regions with rich details. By introducing , the transmittance of these regions can be enhanced, making the restored image clearer in these places; places with larger pixel difference values in the difference image usually correspond to significant changes between the target and the background; by introducing the influence of noise or outliers on the transmittance can be suppressed, making the restored image smoother in these places.

[0082] Use and to normalize the local gradient value and the pixel difference value to ensure that their influence is within a reasonable range and avoid unreasonable deviations of the results caused by too large or too small values.

[0083] The value range of the final transmittance should be restricted within , and truncation processing is performed when it exceeds this range to ensure physical meaning.

[0084] S4: Restore the image content of the corresponding partition of the corresponding polarization image based on the transmittance, and perform weighted fusion and global image enhancement processing on the two restored polarization images according to the polarization degree to obtain the final enhanced image.

[0085] In step S4, the restoration of the image content includes the following steps:

[0086] For each pixel point in each partition, using the optimized perspective rate and background light intensity, the brightness value of the corresponding pixel point in the polarization image is restored through the image restoration formula, where the image restoration formula is:

[0087]

[0088] Where, represents the pixel point The restored brightness value.

[0089] In this embodiment, This operation is to remove the influence of the background light. The background light is the uniform background illumination caused by the scattering of light by water, which will affect the true brightness of the image.

[0090] By subtracting the background light intensity a preliminary image that removes the influence of ambient light is obtained.

[0091] This part further corrects the preliminary result to compensate for the light attenuation caused by water scattering.

[0092] The perspective rate reflects the degree of light loss on the path from the object surface to the camera sensor. A higher perspective rate means less light loss, and a lower perspective rate means more light loss.

[0093] Dividing by is equivalent to "restoring" the light lost due to scattering, making the image look clearer.

[0094] Finally, adding the background light intensity to ensure that the final image still contains the information of ambient light and avoid the image being too dim or losing its natural sense.

[0095] In step S4, the steps of weighted fusion specifically include the following steps:

[0096] Obtain the polarization degree difference value of each pixel point in the two polarization images, and query the corresponding weight factor according to the predefined mapping relation table, where the predefined mapping relation table divides the polarization degree difference value into multiple continuous intervals, and each interval corresponds to a preset weight factor.

[0097] In this embodiment, for the two restored polarized images (0° and 90° polarization direction images), the ratio of the intensity difference between the two to the total intensity is calculated pixel by pixel to obtain a polarization degree difference value ranging from 0 to 1. This value reflects the degree of difference in the polarization characteristics of the two images at the same position: the higher the difference value, the greater the influence of polarization modulation in this area (such as the target edge or high-reflection surface); the lower the difference value, the more similar the polarization characteristics of this area (such as the uniformly scattered background water body).

[0098] The polarization degree difference value is divided into multiple continuous intervals (for example, 10 equal-width intervals, each interval being 0.1), and the weight assignment rules are as follows:

[0099] Low difference interval (0.0 - 0.3): Assign a higher weight to a single image (for example, weight 0.9 for the first image and 0.1 for the second image) to suppress background noise.

[0100] Medium difference interval (0.3 - 0.6): Use progressive weights (for example, 0.3 → 0.7, 0.6 → 0.5) to balance the complementary information of the two images.

[0101] High difference interval (0.6 - 1.0): Distribute weights evenly (for example, 0.5) to retain the detailed features in the dual-polarized image.

[0102] Based on the queried weight factors, the brightness values of the corresponding pixel points in the two restored polarized images are weighted and superimposed to generate a fused image.

[0103] In this embodiment, the weight value of each pixel point is queried according to the predefined mapping table, and the two restored polarized images are weighted and superimposed. Exemplarily, if the polarization degree difference value of a certain pixel is 0.7 (high difference interval), the weights of the two images are both 0.5, and the brightness value of this pixel in the fused image after superposition is the average of the brightness values of the two images.

[0104] If the polarization degree difference value of the target area is significantly higher than that of the background (for example, the difference exceeds 0.5), it is forced to be set to an equal weight (0.5) to avoid detail loss caused by segmentation errors; bilateral filtering (spatial kernel size 5×5, brightness difference threshold 10) is performed on the fused transition area to eliminate edge artifacts caused by weight jumps; dynamic range compression is performed on the fused brightness values to avoid brightness overflow caused by superposition.

[0105] After weighted fusion, global image enhancement is required, including:

[0106] Contrast enhancement: Use limited contrast adaptive histogram equalization, divide the image into 8×8 local grids, and the contrast limit within each grid is 2.0 to enhance details while suppressing noise amplification.

[0107] Color restoration: Extract the chrominance channel of the original polarized image, combine it with the fused luminance channel to restore the natural color of the underwater scene.

[0108] And adjust the gain of the chrominance channel according to the pre-calibrated white balance parameters to correct the blue-green bias caused by the water body.

[0109] Embodiment 2

[0110] The present invention also provides an underwater image enhancement system based on polarization differential imaging for performing an underwater image enhancement method based on polarization differential imaging. As shown in Figure 2 The system includes:

[0111] An image acquisition module 100 for acquiring a depth image corresponding to the underwater image to be enhanced and two polarized images with a polarization degree exceeding a set threshold, and calculating a difference image between the two polarized images;

[0112] A background light intensity calculation module 200 for regionally dividing the two polarized images respectively based on the difference image, and calculating the background light intensity of each divided partition using the depth image;

[0113] A transmittance calculation module 300 for calculating and optimizing the transmittance of each pixel point in the corresponding partition based on the background light intensity.

[0114] An image enhancement module 400 for restoring the image content of the corresponding partition of the corresponding polarized image based on the transmittance, and performing weighted fusion and global image enhancement processing on the two restored polarized images according to the polarization degree to obtain the final enhanced image.

[0115] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also 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 means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0116] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium, which includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0117] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, commodity or device. Without further limitations, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. An underwater image enhancement method based on polarization differential imaging, characterized in that The method includes the following steps: Obtain a depth image corresponding to the underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and calculate a difference image between the two polarization images; Based on the difference image, divide the two polarization images into regions respectively, and calculate the background light intensity of each divided region by using the depth image; Among them, the region division includes: Process the difference image by using a semantic segmentation model based on deep learning, identify and mark the target region, background region and transition region, and at the same time map the marking result to the two polarization images to obtain the target region, background region and transition region of each polarization image; Among them, the calculation of the background light intensity of each region includes: For the background region, obtain the variance of the pixel difference values in the corresponding region of the difference image as a weight factor, and perform weighted average on the pixel values of the depth image to obtain the background light intensity of the background region; For the target region, determine the polarization attenuation coefficient according to the polarization degree ratio of the two polarization images in the target region, combine the local gradient information of the depth image, generate an adaptive weight through a normalized product function, and calculate the background light intensity of the target region; For the transition region, perform edge detection on the difference image and compare it with the pixel intensity difference of the two polarization images to generate a mask, and perform bilinear interpolation on the background light intensity of the background region and the target region based on the mask to obtain the background light intensity of the transition region; Based on the background light intensity, calculate and optimize the transmittance of each pixel point in the corresponding region, where the transmittance represents the effective transmission ratio of light optimized by the background light intensity, gradient value and pixel difference value; Restore the image content of the corresponding region of the corresponding polarization image based on the transmittance, perform weighted fusion on the two restored polarization images according to the polarization degree, and then perform global image enhancement processing on the weighted fusion image to obtain the final enhanced image.

2. The underwater image enhancement method based on polarization difference imaging according to claim 1, wherein The acquisition of the polarization image includes: Use the same imaging device to sequentially obtain two polarization images at different polarization degrees, and the polarization degree difference exceeds the set threshold.

3. The underwater image enhancement method based on polarization difference imaging according to claim 1, wherein The acquisition of the difference image includes: For the two polarization images, perform pixel-level matching at the same spatial position; For each pair of matching pixel points, calculate the difference between the pixel values of the two polarization images and take the absolute value to obtain the difference image.

4. The underwater image enhancement method based on polarization difference imaging according to claim 3, characterized in that, The calculation and optimization of the transmittance of each pixel point in the corresponding region based on the background light intensity includes: For each pixel point in each region, calculate a preliminary transmittance based on the background light intensity and the original brightness value of the pixel point in the polarization image, where the calculation formula is: Among them, represents the background light intensity of the pixel point , and its value range is , represents the original brightness value of the pixel point , represents the preliminary transmittance of the pixel point . Optimize the preliminary transmittance according to the local gradient value of the depth image and the pixel difference value of the difference image to obtain the final transmittance, where the optimization formula is: Among them, and are respectively weight parameters for presetting to control local gradients and the influence of pixel differences, and respectively represent the maximum values of the local gradients of the depth image and the pixel differences of the difference image, and respectively represent the local gradient value and the pixel difference value of the pixel point .

5. The underwater image enhancement method based on polarization difference imaging according to claim 4, characterized in that, The restoration of the image content includes: For each pixel point in each region, use the optimized transmittance and background light intensity to restore the brightness value of the corresponding pixel point in the polarization image through an image restoration formula.

6. The underwater image enhancement method based on polarization difference imaging according to claim 5, wherein The weighted fusion of the two restored polarization images according to the polarization degree includes: Obtain the polarization degree difference value of each pixel point in the two polarization images, and query the corresponding weight factor according to a predefined mapping relation table, where the predefined mapping relation table divides the polarization degree difference value into multiple continuous intervals, and each interval corresponds to a preset weight factor; Based on the queried weight factor, perform weighted superposition on the brightness values of the corresponding pixel points in the two restored polarization images to generate a fused image.

7. An underwater image enhancement system based on polarization differential imaging, which is used to execute an underwater image enhancement method based on polarization differential imaging according to any one of claims 1 to 6, and is characterized in that, The system includes: An image acquisition module, configured to acquire a depth image corresponding to an underwater image to be enhanced and two polarization images with a polarization degree exceeding a set threshold, and calculate a difference image between the two polarization images; A background light intensity calculation module, configured to perform region division on the two polarization images respectively based on the difference image, and calculate the background light intensity of each divided partition by using the depth image; A transmittance calculation module, configured to calculate and optimize the transmittance of each pixel point in the corresponding partition based on the background light intensity, where the transmittance represents the effective transmission ratio of light optimized by the background light intensity, gradient value, and pixel difference value; An image enhancement module, configured to restore the image content of the corresponding partition of the corresponding polarization image based on the transmittance, perform weighted fusion on the two restored polarization images according to the polarization degree, and then perform global image enhancement processing on the weighted fused image to obtain a final enhanced image.

Images