An underwater polarization imaging method based on a micro-polarizer array

Abstract

The embodiment of the application discloses a kind of underwater polarization imaging methods based on micro polaroid array, it is related to optical imaging technical field, the method comprises: obtaining the polarization image of same underwater scene in four polarization directions;Based on four images, calculate linear polarization degree, and construct confidence map;Four images are constituted into two pairs of orthogonal images;For each pair of orthogonal images, one of the images is contrast enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement;Based on two pairs of enhanced orthogonal images respectively, image restoration is carried out using estimated backscattering polarization degree map and backscattering upper limit value;Based on the preset local quality evaluation index, two restoration results are adaptively weighted and fused, and the target underwater restoration image is output;The method can realize fast, stable, high-contrast and polarization physical consistency in strong scattering, high turbidity underwater environment Image restoration.

Description

Technical Field

[0001] This invention relates to the field of optical imaging technology, and in particular to an underwater polarization imaging method based on a micro polarizer array. Background Technology

[0002] The degradation mechanism of underwater optical imaging mainly stems from scattering and absorption: the reflected or radiated light from the target is attenuated during propagation, reducing the direct component; simultaneously, backscattering from water particles enters the detector, forming a hazy background, thus significantly reducing image contrast. Classical imaging models for turbid media typically express the detector's received irradiance as the superposition of the direct and backscattered components, approximating the upper bound of backscattering as a global constant. Based on this, image restoration can be achieved by estimating backscattering and medium transmittance.

[0003] Building upon this foundation, polarization imaging is widely used to improve visibility in turbid media imaging scenarios. Its core idea lies in the difference in polarization characteristics between backscattered light and the target's direct light. By employing linearly polarized illumination and combining it with a polarizer to acquire two orthogonal polarized images, parameters such as the degree of backscattering polarization and the upper limit of backscattering can be estimated, thereby restoring the target's radiance. In classic dehazing and restoration studies, Schechner et al. utilized polarization information to acquire orthogonal images and perform parameter estimation, providing an important reference for polarization restoration direction. Simultaneously, the effectiveness of underwater polarization contrast enhancement and target detection has been verified by numerous experimental studies; for example, acquiring two orthogonal polarization states can improve target visibility and suppress scattering background.

[0004] However, traditional polarization restoration often relies on frame-by-frame acquisition using a mechanically rotating analyzer, or at least a two-frame orthogonal polarization acquisition method. In dynamic scenes, these acquisition strategies are prone to introducing time asynchrony and motion artifacts, which is detrimental to system miniaturization and real-time applications. Furthermore, under conditions of strong scattering or high turbidity, although polarization filtering can suppress backscattering to some extent, the grayscale dynamic range of the original orthogonal polarization image may still be compressed into a narrow range, leading to unstable subsequent parameter estimation and problems such as local over-amplification in the restoration result.

[0005] Therefore, there is an urgent need for a new method that can closely integrate the hardware characteristics of micro-polarizer arrays, make full use of multi-directional polarization information, and achieve stable, high-contrast, and physically consistent underwater image restoration under strong scattering environments. Summary of the Invention

[0006] The main objective of this invention is to provide an underwater polarization imaging method based on a micro-polarizer array, which can achieve high visibility, high contrast, and polarization-physical consistency restoration results in polarization imaging restoration scenarios facing strong underwater scattering environments.

[0007] To achieve the above objectives, the first aspect of this application provides an underwater polarization imaging method based on a micro-polarizer array, the method comprising: Acquire polarization images of the same underwater scene in four polarization directions; The linear polarization degree is calculated based on the image of the four polarization directions, and a confidence map is constructed based on the linear polarization degree. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image. The images of the four polarization directions are configured into two pairs of orthogonal images. For each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images. Using the confidence plot as a constraint, the spatially adaptive backscattering polarization plot and the upper limit of backscattering are estimated; Based on two pairs of enhanced orthogonal images, image restoration is performed using the estimated backscatter polarization degree map and the backscatter upper limit value to obtain two restoration results. Based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the target underwater restored image.

[0008] A second aspect of this application provides an underwater polarization imaging device based on a micro-polarizer array, comprising: The image acquisition module is used to acquire polarization images of the same underwater scene in four polarization directions; The polarization information processing module is used to calculate the degree of linear polarization based on the image in the four polarization directions, and to construct a confidence map based on the degree of linear polarization. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image. The image enhancement module is used to form two pairs of orthogonal images from the four polarization directions; for each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images. The parameter estimation module is used to estimate the spatially adaptive backscattering polarization map and the upper limit of backscattering using the confidence map as a constraint; The image restoration and fusion module is used to restore the image based on two pairs of enhanced orthogonal images, using the estimated backscatter polarization degree map and the backscatter upper limit value, to obtain two restoration results; based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the target underwater restored image.

[0009] A third aspect of this application provides an electronic device including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform steps as described in the first aspect and any possible implementation thereof.

[0010] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the first aspect and any possible implementation thereof.

[0011] This application provides an underwater polarization imaging method based on a micro-polarizer array. The method acquires polarization images of the same underwater scene in four polarization directions. Linear polarization degrees are calculated based on the images in the four polarization directions, and a confidence map is constructed based on these degrees. The confidence map characterizes the reliability of polarization information for each pixel in the image. The images in the four polarization directions are configured into two pairs of orthogonal images. For each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement to maintain consistent polarization relationships between the two enhanced images. Using the confidence map as a constraint, a spatially adaptive backscattering polarization degree map and a backscattering upper limit value are estimated. Based on the two pairs of enhanced orthogonal images, image restoration is performed using the estimated backscattering polarization degree map and the backscattering upper limit value, resulting in two restoration results. Based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output a target underwater restored image.

[0012] The technical solution provided in this application has the following beneficial effects: By acquiring four-directional polarization images and constructing polarization confidence maps, spatial differentiation of polarization information reliability was achieved, providing a reliable basis for subsequent processing. An image processing method that enhances only one image and reconstructs another based on polarization constraints effectively expands the dynamic range of the images while strictly maintaining the physical consistency between orthogonal polarization images, avoiding distortion of polarization relationships. Spatially adaptive backscattering parameter estimation using confidence map constraints significantly reduces anomaly interference, improves the stability of parameter estimation, and effectively suppresses over-magnification in distant regions. Finally, adaptive fusion of the restoration results from the two orthogonal polarization channels fully utilizes the complementarity of information from different polarization directions, significantly improving the overall contrast, detail clarity, robustness, and visual quality of the final restored image in complex underwater scenarios. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] in: Figure 1 This is a schematic flowchart of an underwater polarization imaging method based on a micro-polarizer array provided in an embodiment of this application. Figure 2 This is a schematic diagram of a polarization image and its restoration result provided in an embodiment of this application; Figure 3 This is a schematic diagram of another polarization image and restoration result provided in an embodiment of this application; Figure 4 This is a schematic diagram illustrating the response changes of four types of oriented pixels under a rotating polarizer condition, as provided in an embodiment of this application. Figure 5A This is a schematic diagram of the EME incremental statistical results of experimental data provided in an embodiment of this application; Figure 5B This is a schematic diagram of the entropy increment statistics of experimental data provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0016] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0017] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0018] The Stokes parameters mentioned in the embodiments of this application refer to the polarization state of a beam of light completely characterized by four real numbers S0, S1, S2, and S3. In imaging scenarios that only consider linear polarization, the first three can be used, where S0 is the total light intensity, S1 is the light intensity difference between the 0° and 90° directions, and S2 is the light intensity difference between the 45° and 135° directions.

[0019] The embodiments of this application are described below with reference to the accompanying drawings.

[0020] This application proposes an underwater polarization image restoration method based on micropolarizer array data. Addressing typical underwater degradation conditions such as strong scattering and high turbidity, it utilizes multi-directional polarization observation information to calculate polarization parameters and completes image enhancement and physical model restoration while maintaining consistent polarization relationships. Compared to existing methods that primarily rely on two orthogonal polarization images with relatively fixed enhancement methods, this application introduces four-directional Stokes calculations, confidence constraints based on polarization degree, spatially adaptive scattering parameter estimation, and fusion of biorthogonal pair results to improve the restoration effect and applicability in complex underwater scenarios.

[0021] Figure 1 A flowchart illustrating an underwater polarization imaging method based on a micro-polarizer array, as provided in this application embodiment, is shown below. Figure 1 As shown, the method includes: 101. Obtain polarization images of the same underwater scene in four polarization directions.

[0022] The method in this application embodiment can be implemented by an underwater polarization imaging device based on a micro polarizer array, which can be implemented in an electronic device such as a terminal device in practical applications.

[0023] This application can use four polarization images as input. The experimental data can be adjusted as needed, for example, by selecting a suitable and reliable data source; each set of data can include grayscale polarization images of the same scene in four polarization directions, such as 0°, 45°, 90°, and 135°, with the corresponding image data denoted as follows: I 0、 I 45 , I 90 ,I 135 This application can directly perform subsequent calculations and restoration processing based on the above four images.

[0024] 102. Calculate the linear polarization degree based on the images of the above four polarization directions, and construct a confidence map based on the above linear polarization degree. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image.

[0025] Specifically, polarization information can be calculated from images in four polarization directions, including Stokes parameters, degree of linear polarization, and angle of linear polarization (AoLP).

[0026] Calculate the degree of linear polarization and the polarization angle:

[0027] in To prevent tiny constants with a denominator of zero.

[0028] Furthermore, confidence graphs can be constructed based on DoLP. When the scene contains areas with fragmented textures or significant local noise, a low-gradient mask can be overlaid on the confidence map to reduce the impact of these areas on subsequent parameter estimation. The confidence map can be used to guide backscattering parameter estimation, fusion weight allocation, and parameter selection.

[0029] In one alternative implementation, the confidence plot is obtained by cropping the values ​​of the linear polarization degree to the [0,1] interval.

[0030] Unlike conventional methods that primarily process two orthogonally polarized images, this application utilizes four polarization images (0°, 45°, 90°, and 135°) to calculate Stokes parameters and further obtain the linear polarization degree (DoLP) and polarization angle (AoLP). Simultaneously, a confidence map C=clip(DoLP,0,1) is constructed using DoLP to characterize the reliability of polarization information for each pixel, and low-gradient suppression can be superimposed to reduce unstable estimations in textured areas. This approach spatially distinguishes between "reliable polarization information regions" and "unreliable polarization information regions," providing a basis for subsequent parameter estimation and restoration. This reduces the interference of noise and texture on polarization estimation, improving the stability and consistency of the algorithm in complex underwater scenarios.

[0031] 103. The images of the above four polarization directions are configured into two pairs of orthogonal images; for each pair of the above orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images.

[0032] In this embodiment, only linear histogram stretching and polarization constraint reconstruction of an orthogonal polarization image are performed.

[0033] The contrast enhancement mentioned above can be achieved by stretching a linear histogram, specifically including: The grayscale values ​​of the selected single orthogonal polarized image are linearly mapped from the original dynamic range to the preset target dynamic range.

[0034] Optionally, the images of the above four polarization directions are configured into two pairs of orthogonal images, including: The images of the above four polarization directions are configured into two pairs of orthogonal images: (0°, 90°) and (45°, 135°).

[0035] Specifically, taking the four polarization directions of 0°, 45°, 90°, and 135° as examples, the four polarization images can be used to construct two pairs of positive polarization channels: ( I 0, I 90 )and( I 45 , I 135 ); For each pair of orthogonal polarization channels, select one orthogonal polarization image and perform linear histogram stretching, adjusting its grayscale dynamic range from [I min I max Linear mapping is applied to a preset range to expand contrast and alleviate grayscale compression. The other positive traffic channel is not stretched separately, but is reconstructed based on the polarization constraint relationship between the two positive traffic channels, so that the polarization relationship is consistent before and after processing (e.g., keeping the polarization degree unchanged), thereby avoiding the destruction of polarization relationship and distortion of polarization parameters caused by stretching separately.

[0036] 104. Using the confidence plot above as a constraint, estimate the spatially adaptive backscattering polarization plot and the upper limit of backscattering.

[0037] Specifically, the first step is to provide an initial backscattering amount for subsequent estimation. According to the principle of polarization imaging, the backscattering amount can be approximately estimated by half the sum of two orthogonal polarization images.

[0038] In one alternative implementation, the above-mentioned estimation of the spatially adaptive backscattered polarization degree map includes: In the high confidence and low gradient region indicated by the confidence map, the high quantile of the polarization degree is calculated as the initial estimate of the local backscattering polarization degree. The high confidence level indicates that the confidence value in the confidence map is greater than the first confidence threshold, and the low gradient level indicates that the gradient value in the confidence map is less than the first gradient threshold. The initial estimate is propagated to the entire image through filtering to obtain the spatially adaptive backscattering polarization map.

[0039] Specifically, to adapt to the spatial non-uniformity of underwater scattering, embodiments of this application introduce a spatially variable backscattering polarization degree p. A The estimation method for (x,y) is as follows: In regions with high confidence C and low local gradients, the high quantile of polarization degree P (e.g., 90%–95%) is used as the value of that region. The value is then propagated across the entire graph using guided filtering or bilateral filtering to obtain p. A (x, y). When the local confidence level is low, it can degenerate into a global mean method to maintain estimation stability. The high-confidence region can be understood as the set of pixels in the confidence map whose confidence value is greater than a first confidence threshold, and the low-gradient region can be understood as the set of pixels in the polarization image whose gradient magnitude is less than a first gradient threshold. These thresholds can be set as needed. In an optional implementation, the estimation of the backscattering upper limit using the confidence map as a constraint specifically includes: In the confidence map above, pixels with confidence values ​​greater than a preset second confidence threshold are selected to form a target pixel set; Calculate the quantiles between the preset proportions of the backscattering amount corresponding to the above target pixel set, and use them as the initial estimate of the above backscattering upper limit value; The initial estimate was fine-tuned using a preset safety factor to obtain the final upper limit value for backscattering.

[0040] Specifically, the aforementioned preset ratio range can be set as needed. That is, in regions with high confidence and low local gradients, the high quantile of polarization degree P (e.g., 90%–95%) is used as the local gradient. The estimate is obtained, and p is propagated across the entire graph using guided filtering or bilateral filtering. A (x,y). Simultaneously, an upper limit for backscattering can be set. Introducing confidence masks and quantile estimation: When satisfying... On the pixel set, take the high quantile (e.g., 99%) of the backscattering amount A as... The estimate can be made and a preset safety factor can be used. (For example Fine-tuning was performed on it to reduce the impact of extreme points on the recovery process.

[0041] The advantage of this approach is that quantile statistics and confidence masks can effectively suppress outliers, glare, vignetting, and local noise. Interference; Local adaptive p A(x,y) can better adapt to the non-uniform scattering background in space, making the estimation of transmittance t and restored radiance L more reasonable, significantly reducing long-distance over-amplification and noise amplification, and improving restoration consistency.

[0042] 105. Based on two pairs of enhanced orthogonal images, image restoration is performed using the estimated backscatter polarization degree map and the backscatter upper limit value to obtain two restoration results. Based on the preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the underwater restored image of the target.

[0043] In one optional implementation, the aforementioned preset local quality evaluation index is a weighted sum of local contrast measurement, image gradient information, and confidence map.

[0044] Specifically, this can be based on the estimated backscattering amount. A With the upper limit of backscattering Calculate transmittance:

[0045] This yields the restored image L (represented using a common polarization restoration model):

[0046] in This represents the intensity term after the aforementioned polarization-consistent enhancement. To avoid numerical instability, a lower limit threshold is set for t, and the output dynamic range is constrained.

[0047] The restoration results were obtained based on two pairs of orthogonal traffic paths: (0° / 90°) and (45° / 135°). and Calculate the quality weight of each pair of results in the local region, and normalize to obtain the weight. w 1. w 2. And then merge them:

[0048]

[0049] This fusion can improve the stability and consistency of the overall restoration by utilizing the effective information from another pair of channels when one pair of channels is weakly affected by noise or local polarization.

[0050] Finally, the restored image can be output. L final It also outputs the EME increment and entropy increment of 17 sets of experimental images.

[0051] The core of the method in this application is to construct polarization parameters and confidence constraints based on four-directional polarization observations, and to achieve enhancement, scattering parameter estimation and restoration under the premise of consistent polarization relationship. At the same time, the method is combined with adaptive fusion of two pairs of positive scattering channels to improve the applicability of the method under complex underwater scattering conditions.

[0052] To verify the effectiveness of this technical solution, experimental verification examples are given below.

[0053] This application uses underwater polarization image data provided by a collaborating team as input. Each set of data contains grayscale polarization images of the same scene at four polarization directions: 0°, 45°, 90°, and 135°. Based on the four-directional polarization images, the algorithm calculates the Stokes parameters and obtains the linear polarization degree DoLP and polarization angle AoLP, and then constructs a confidence map to constrain the estimation of backscattering-related parameters.

[0054] Subsequently, for the two pairs of orthogonal traffic channels (0° / 90° and (45° / 135°), linear histogram stretching is applied to one image to expand the grayscale dynamic range, and the other orthogonal traffic channel is reconstructed while maintaining consistent polarization relationships. Finally, the restored results of the two pairs of orthogonal traffic channels are obtained separately, and then fused using spatial adaptive weights to output the final restored image.

[0055] Figure 2 and Figure 3 This is an example of the results from selecting a portion of the experimental data, such as... Figure 2 and Figure 3 As shown, the left side represents the restoration result of this application's scheme, and the right side represents the four corresponding original polarization images. Figure 2 and Figure 3 As can be seen, the original four-directional polarization image has low overall contrast and unclear target details under strong scattering conditions; after processing by the technical solution of this application, the restoration result is enhanced in terms of target outline and local details, while the background fogging effect is reduced.

[0056] Figure 4 The response variations of four types of oriented pixels under a rotating polarizer condition are illustrated. As the polarizer angle changes, the average response of different oriented channels exhibits a periodic variation that approximately conforms to Malus's law, and shows phase shifts between different orientations. This phenomenon characterizes the difference in response of different oriented channels to linear polarization directions, helping to explain the consistency between the orientation correspondence of the four polarization channels and the signal, thus supporting subsequent calculations based on polarization constraints.

[0057] To further provide a quantitative characterization, this application statistically analyzed the changing trends of local contrast and information content before and after restoration for 17 sets of experimental data. Figure 5AThe statistical results of EME increments for 17 sets of experimental data are presented. EME is used to characterize the local contrast level of an image, and a positive EME increment indicates an improvement in local contrast after restoration. Figure 5B The statistical results of entropy increments for 17 sets of experimental data are presented. Entropy is used to characterize the information content of gray-level distribution; a positive entropy increment indicates an increase in gray-level levels and information content after restoration. Figure 5A and Figure 5B It is evident that the technical solution of this application demonstrates a trend of increased contrast and information content across multiple sets of data from different scenarios.

[0058] Based on the above embodiments, this application utilizes four-directional polarization images to calculate Stokes parameters and further obtains the linear polarization degree DoLP and polarization angle AoLP. Simultaneously, a confidence map is constructed using DoLP to indicate reliable regions of polarization information. This technique can spatially distinguish between regions with more reliable and less reliable polarization information, reducing the impact of fine texture areas and local noise on polarization parameter calculations, and providing a more reasonable basis for subsequent scattering parameter estimation and image restoration.

[0059] The method in this application employs a process of "linear histogram stretching of only one orthogonal polarization image and reconstructing another orthogonal polarization path under polarization degree constraints" in the contrast enhancement stage, ensuring that the polarization relationship between the orthogonal polarization images remains consistent before and after enhancement. This technique can alleviate the problem of grayscale dynamic range compression under strong scattering conditions, while avoiding the destruction of polarization relationship and distortion of polarization parameters caused by enhancing two orthogonal images separately, thus contributing to a more natural restoration result.

[0060] The method in this application introduces a confidence mask and quantile statistics in the estimation of backscattering polarization degree and backscattering upper limit, and combines spatial propagation to obtain spatially adaptive estimation results. This technique can reduce the interference of anomalous factors such as reflection points, vignetting, and local noise on the estimation of scattering parameters, and effectively suppress the over-amplification phenomenon that may occur in distant regions, making the restoration process more consistent with the spatial non-uniformity of underwater scattering.

[0061] The method in this application obtains restoration results based on two pairs of orthogonal channels and performs adaptive fusion through weighting. This technique can compensate for the effects of noise, saturation, or weak local polarization on one pair of channels by utilizing the effective information of the other pair of channels, thereby reducing the impact of local artifacts and unstable regions on the results and improving the overall consistency and appearance of the restored image.

[0062] In terms of experimental verification, this application performs restoration processing on multiple sets of underwater polarization data and outputs the EME increment and entropy increment of the restored image to characterize the improvement in image contrast and information content. The EME is used to measure the degree of local contrast enhancement; a larger EME generally indicates higher contrast. The above experimental results can serve as quantitative evidence of the improved restoration effect of this application.

[0063] Based on the description of the foregoing method embodiments, this application also proposes an underwater polarization imaging device based on a micro-polarizer array. The device includes: The image acquisition module is used to acquire polarization images of the same underwater scene in four polarization directions; The polarization information processing module is used to calculate the degree of linear polarization based on the image in the four polarization directions, and to construct a confidence map based on the degree of linear polarization. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image. The image enhancement module is used to form two pairs of orthogonal images from the four polarization directions; for each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images. The parameter estimation module is used to estimate the spatially adaptive backscattering polarization map and the upper limit of backscattering using the confidence map as a constraint; The image restoration and fusion module is used to restore the image based on two pairs of enhanced orthogonal images, using the estimated backscatter polarization degree map and the backscatter upper limit value, to obtain two restoration results; based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the target underwater restored image.

[0064] In one embodiment of this application, an electronic device is also provided. See also... Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6 As shown, the electronic device 700 includes a processor 701 and a memory 702. The memory 702 stores a computer program, which, when executed by the processor 701, will perform actions such as... Figure 1 Any step in the method embodiment shown. The electronic device 700 may also include input / output devices, etc. In a specific embodiment, the electronic device may be a terminal device, etc.

[0065] In one embodiment, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor 701, causes the processor 701 to perform any of the steps in the above method embodiments.

[0066] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0067] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0068] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

Claims

1. An underwater polarization imaging method based on a micro-polarizer array, characterized in that, The method includes: Acquire polarization images of the same underwater scene in four polarization directions; The linear polarization degree is calculated based on the image of the four polarization directions, and a confidence map is constructed based on the linear polarization degree. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image. The images of the four polarization directions are configured into two pairs of orthogonal images. For each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images. Using the confidence plot as a constraint, the spatially adaptive backscattering polarization plot and the upper limit of backscattering are estimated; Based on two pairs of enhanced orthogonal images, image restoration is performed using the estimated backscatter polarization degree map and the backscatter upper limit value to obtain two restoration results. Based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the target underwater restored image.

2. The underwater polarization imaging method based on a micro-polarizer array according to claim 1, characterized in that, The four polarization directions include 0°, 45°, 90°, and 135°. The step of constructing two pairs of orthogonal images from the four polarization directions includes: The images of the four polarization directions are configured into two pairs of orthogonal images: (0°, 90°) and (45°, 135°).

3. The underwater polarization imaging method based on a micro-polarizer array according to claim 1, characterized in that, The confidence plot is obtained by cropping the values ​​of the linear polarization degree to the [0,1] interval.

4. The underwater polarization imaging method based on a micro-polarizer array according to claim 1 or 3, characterized in that, The estimated spatially adaptive backscattering polarization map includes: In the high-confidence and low-gradient region indicated by the confidence map, the high quantile of the polarization degree is calculated as an initial estimate of the local backscattering polarization degree. The high confidence level indicates that the confidence value in the confidence map is greater than a first confidence threshold, and the low gradient level indicates that the gradient value in the confidence map is less than a first gradient threshold. The initial estimate is propagated to the entire image through a filtering operation to obtain the spatially adaptive backscattering polarization map.

5. The underwater polarization imaging method based on a micro-polarizer array according to claim 1, characterized in that, The preset local quality evaluation index is a weighted sum of local contrast measurement, image gradient information, and the confidence map.

6. The underwater polarization imaging method based on a micro-polarizer array according to claim 1, characterized in that, The contrast enhancement is a linear histogram stretching, specifically including: The grayscale values ​​of the selected single orthogonal polarized image are linearly mapped from the original dynamic range to the preset target dynamic range.

7. The underwater polarization imaging method based on a micro-polarizer array according to claim 1, characterized in that, The step of using the confidence map as a constraint to estimate the upper limit of backscattering specifically includes: In the confidence map, pixels with confidence values ​​greater than a preset second confidence threshold are selected to form a target pixel set; Calculate the quantiles between the preset proportions of the backscattering amount corresponding to the target pixel set, and use them as the initial estimate of the upper limit of the backscattering; The initial estimate is fine-tuned using a preset safety factor to obtain the final upper limit value for backscattering.

8. An underwater polarization imaging device based on a micro-polarizer array, characterized in that, include: The image acquisition module is used to acquire polarization images of the same underwater scene in four polarization directions; The polarization information processing module is used to calculate the degree of linear polarization based on the image in the four polarization directions, and to construct a confidence map based on the degree of linear polarization. The confidence map is used to characterize the reliability of the polarization information of each pixel in the image. The image enhancement module is used to form two pairs of orthogonal images from the four polarization directions; for each pair of orthogonal images, only one image is contrast-enhanced, and the other image is reconstructed based on the polarization degree constraint relationship before and after enhancement, so as to maintain the consistency of the polarization relationship between the two enhanced images. The parameter estimation module is used to estimate the spatially adaptive backscattering polarization map and the upper limit of backscattering using the confidence map as a constraint; The image restoration and fusion module is used to restore the image based on two pairs of enhanced orthogonal images, using the estimated backscatter polarization degree map and the backscatter upper limit value, to obtain two restoration results. Based on a preset local quality evaluation index, the two restoration results are adaptively weighted and fused to output the target underwater restoration image.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, the processor performs the steps of the method as described in any one of claims 1-7.

Images