Sonar Image Optimization Method, System, Device and Storage Medium

Abstract

The present application discloses a sonar image optimization method, system, device and storage medium. The method sequentially performs double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction and Hough line removal processing on the initial two-dimensional sonar image. Among them, the double-threshold segmentation enhancement processing can highlight the essential features of the initial two-dimensional sonar image to exclude the interference of system factors; the skeletonization processing helps to extract linear features and improve the accuracy of subsequent data analysis; the Canny edge feature extraction can better ensure the integrity of the edge features of the image; the Hough line removal can specifically remove the irrelevant line segments in the image, thereby reducing the subsequent computational complexity. Therefore, the above image processing steps can effectively ensure the integrity of the two-dimensional sonar image features, and further ensure the reliability of the sonar three-dimensional point cloud data.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium batteries, and in particular, to a method, system, device and storage medium for optimizing sonar images. Background Art

[0002] In a complex underwater environment, the illuminance is low, and the detection range of optical devices is greatly limited. Under the same conditions, acoustic devices perceive the environment by transmitting and receiving ultrasonic waves. They do not rely on light, are not easily interfered with, and have a far detection range, making them more suitable for monitoring complex underwater environments. Using a multi-beam sonar for target detection and three-dimensional reconstruction of underwater structures is very challenging due to the influence of sonar noise and multi-path acoustic reflections, and it cannot intuitively display the underwater environment. For this reason, experts and scholars have conducted research on underwater target detection and three-dimensional reconstruction by using machine learning and image processing technologies.

[0003] However, using the machine learning method requires a dataset containing a large number of real sonar image samples to train the network model so that the model can detect structural targets from the images. Structures that are not trained or even similar structures are very likely to not be detected by the network. In the process of image processing in the prior art, relying only on conventional processing methods may result in too much clutter, long processing time, and unclear two-dimensional image features, resulting in missing, redundant, or misaligned point clouds.

[0004] Therefore, in the process of sonar image processing in the prior art, there is a problem of low reliability of image features. Summary of the Invention

[0005] In view of this, it is necessary to provide a method, system, device and storage medium for optimizing sonar images to solve the problem of low reliability of image features in the process of sonar image processing in the prior art.

[0006] To solve the above problems, the present invention provides a method for optimizing sonar images, including:

[0007] Obtain an initial two-dimensional sonar image;

[0008] Perform double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image in sequence to determine a filtered two-dimensional sonar image;

[0009] Perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

[0010] Further, performing double-threshold segmentation enhancement processing on the initial two-dimensional sonar image includes:

[0011] Set a mask for the initial two-dimensional sonar image, and obtain the effective image pixels of the initial two-dimensional sonar image based on the mask;

[0012] Set a brightness threshold, and determine the bright image set in the initial two-dimensional sonar image corresponding to the effective image pixels according to the brightness threshold;

[0013] Set a mapping threshold in the bright image set, determine a gray mapping function based on the mapping threshold, and perform mapping processing on the effective image pixels according to the gray mapping function to obtain a two-dimensional enhanced sonar image.

[0014] Further, perform skeletonization processing on the initial two-dimensional sonar image, including:

[0015] Perform binarization processing on the pixels of the two-dimensional enhanced sonar image to obtain a binarized image;

[0016] Perform skeleton line calculation processing on the binarized image to obtain a two-dimensional skeletonized sonar image.

[0017] Further, perform Canny edge feature extraction on the initial two-dimensional sonar image, including:

[0018] Perform edge extraction processing on the two-dimensional skeletonized sonar image through a Canny operator to highlight the edge features of the two-dimensional enhanced sonar image and obtain a two-dimensional edge sonar image.

[0019] Further, perform Hough line elimination processing on the initial two-dimensional sonar image, including:

[0020] Perform probabilistic Hough line processing on the two-dimensional edge sonar image to obtain a line segment set of the two-dimensional edge sonar image;

[0021] Set a first line segment length threshold, a second line segment length threshold, and a feature line segment;

[0022] Perform elimination processing on the line segment set according to the first line segment length threshold, the second line segment length threshold, the feature line segment, and a preset line segment elimination relationship to obtain a filtered two-dimensional sonar image.

[0023] Further, the preset line segment elimination relationship is:

[0024]

[0025] Among them, Length L is the length of line segment L, Q1 is the first line segment length threshold, Q2 is the second line segment length threshold, & represents the AND relationship, L1, L2, L3 are the feature line segments of line segment L, means existence, and l is any one of L1, L2, L3.

[0026] Further, perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data, including:

[0027] Perform spatial transformation, filtering processing, downsampling processing, and point cloud fusion processing on the filtered two-dimensional sonar image in sequence to obtain sonar three-dimensional point cloud data.

[0028] To solve the above problems, the present invention also provides a sonar image optimization system, including:

[0029] An initial two-dimensional sonar image acquisition module, configured to acquire an initial two-dimensional sonar image;

[0030] A filtered two-dimensional sonar image determination module, configured to perform double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image in sequence to determine the filtered two-dimensional sonar image;

[0031] A sonar three-dimensional point cloud data acquisition module, configured to perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

[0032] To solve the above problems, the present invention also provides a sonar image optimization device, including a memory and a processor, wherein,

[0033] The memory is used to store programs;

[0034] The processor is coupled to the memory and is configured to execute the programs stored in the memory to implement the steps in the sonar image optimization method, system, device, and storage medium as described above.

[0035] To solve the above problems, the present invention also provides a computer-readable storage medium, configured to store computer-readable programs or instructions, and when the programs or instructions are executed by a processor, they can implement the steps in the sonar image optimization method, system, device, and storage medium as described above.

[0036] The beneficial effects of adopting the above embodiments are as follows: The present invention provides a sonar image optimization method, system, device, and storage medium. This method performs double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image in sequence. Among them, the double-threshold segmentation enhancement processing can highlight the essential features of the initial two-dimensional sonar image to exclude the interference of system factors; the skeletonization processing helps to extract linear features and improve the accuracy of subsequent data analysis; the Canny edge feature extraction can better ensure the integrity of the edge features of the image; the Hough line removal can specifically remove the irrelevant line segments in the image, thereby reducing the subsequent calculation complexity. Therefore, the above image processing steps can effectively ensure the integrity of the two-dimensional sonar image features, and further ensure the reliability of the sonar three-dimensional point cloud data. Description of the Drawings

[0037] Figure 1 It is a schematic flowchart of an embodiment of the sonar image optimization method provided by the present invention;

[0038] Figure 2 It is a schematic flowchart of an embodiment of the double-threshold segmentation enhancement processing for the initial two-dimensional sonar image provided by the present invention;

[0039] Figure 3 It is a schematic flowchart of an embodiment of the Hough line removal processing for the initial two-dimensional sonar image provided by the present invention;

[0040] Figure 4 It is a structural block diagram of an embodiment of the sonar image optimization system provided by the present invention;

[0041] Figure 5 It is a structural block diagram of an embodiment of the sonar image optimization device provided by the present invention. Detailed Embodiments

[0042] The following will specifically describe the preferred embodiments of the present invention in conjunction with the drawings. The drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0043] Before presenting the embodiments, the meaning of OpenCV is explained first:

[0044] OpenCV is a cross-platform computer vision and machine learning software library distributed under the BSD license (open source). It can run on Linux, Windows, Android, and Mac OS operating systems. It is lightweight and efficient - composed of a series of C functions and a small number of C++ classes, and at the same time provides interfaces for languages such as Python, Ruby, and MATLAB, implementing many general algorithms in image processing and computer vision, including a series of basic methods in image processing. Some methods can be directly used. In addition, we can also perform a series of optimization operations based on the native code and interfaces of OpenCV.

[0045] In a complex underwater environment, the light intensity is low, and the detection range of optical devices is greatly limited. Under the same conditions, acoustic devices perceive the environment based on transmitting and receiving ultrasonic waves. They do not rely on light, are not easily interfered with, and have a far detection range, making them more suitable for monitoring complex underwater environments. Using a multi-beam sonar for target detection and three-dimensional reconstruction of underwater structures is very challenging due to the influence of sonar noise and multi-path acoustic reflections, and it cannot intuitively display the underwater environment. Therefore, experts and scholars have conducted research on underwater target detection and three-dimensional reconstruction by using machine learning and image processing techniques.

[0046] However, using the machine learning method requires a dataset containing a large number of real sonar image samples to train the network model so that the model can detect structural targets from the images. Untrained structures, or even similar structures, are very likely to not be detected by the network. In the process of image processing in the prior art, relying only on conventional processing methods may result in excessive clutter, long processing time, and unclear two-dimensional image features, resulting in missing, redundant, or misaligned point clouds.

[0047] Therefore, in the process of sonar image processing in the prior art, there is a problem of low reliability of image features.

[0048] To solve the above problems, the present invention provides a sonar image optimization method, system, device, and storage medium, which will be described in detail below.

[0049] Figure 1 As shown in the flowchart of an embodiment of the sonar image optimization method provided by the present invention, Figure 1 the sonar image optimization method includes:

[0050] Step S101: Obtain an initial two-dimensional sonar image;

[0051] Step S102: Perform double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image in sequence to determine a filtered two-dimensional sonar image;

[0052] Step S103: Perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

[0053] In this embodiment, first, an initial two-dimensional sonar image is obtained; second, double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing are performed on the initial two-dimensional sonar image in sequence to determine a filtered two-dimensional sonar image; finally, three-dimensional reconstruction is performed on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

[0054] In this embodiment, the initial two-dimensional sonar image is sequentially subjected to double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line elimination processing. Among them, the double-threshold segmentation enhancement processing can highlight the essential features of the initial two-dimensional sonar image to exclude the interference of system factors; the skeletonization processing helps to extract linear features and improve the accuracy of subsequent data analysis; the Canny edge feature extraction can better ensure the integrity of the edge features of the image; the Hough line elimination can specifically remove the irrelevant line segments in the image, thereby reducing the subsequent calculation complexity. Therefore, the above image processing steps can effectively ensure the integrity of the two-dimensional sonar image features, and further ensure the reliability of the sonar three-dimensional point cloud data.

[0055] It should be noted that the above sonar image optimization method can be implemented based on the relevant functions in the open-source OpenCV technology, which will not be elaborated here.

[0056] In other embodiments, the technical platform for implementing the above sonar image optimization method can also be adjusted according to actual needs, which is not limited here.

[0057] As a preferred embodiment, in step S102, in order to perform double-threshold segmentation enhancement processing on the initial two-dimensional sonar image, as Figure 2 shown, Figure 2 is a schematic flowchart of an embodiment for performing double-threshold segmentation enhancement processing on the initial two-dimensional sonar image provided by the present invention, including:

[0058] Step S121: Set a mask for the initial two-dimensional sonar image, and obtain the effective image pixels of the initial two-dimensional sonar image based on the mask;

[0059] Step S122: Set a brightness threshold, and determine the bright image set in the initial two-dimensional sonar image corresponding to the effective image pixels according to the brightness threshold;

[0060] Step S123: Set a mapping threshold in the bright image set, determine a gray mapping function based on the mapping threshold, and perform mapping processing on the effective image pixels according to the gray mapping function to obtain a two-dimensional enhanced sonar image.

[0061] In this embodiment, first, a mask is set for the initial two-dimensional sonar image, and the effective image pixels of the initial two-dimensional sonar image are obtained based on the mask; then, a brightness threshold is set, and the bright image set in the initial two-dimensional sonar image corresponding to the effective image pixels is determined according to the brightness threshold; finally, a mapping threshold in the bright image set is set, a gray mapping function is determined based on the mapping threshold, and mapping processing is performed on the effective image pixels according to the gray mapping function to obtain a two-dimensional enhanced sonar image.

[0062] In this embodiment, pixels in the initial two-dimensional sonar image are divided into valid pixels and invalid pixels through a mask, and two thresholds are defined. Double-threshold segmentation enhancement processing is performed on the pixels within the mask. Finally, gray mapping is performed to map pixel values of different brightness levels to new brightness levels, thereby enhancing the contrast and features of the image.

[0063] In a specific embodiment, the two thresholds are specifically:

[0064] Design the percentile Pe1 of the pixels in the denoised image as the brightness threshold λ p , and by checking the value of λ p , it can be determined whether the current image is a bright image;

[0065] Take the set M of all pixel values greater than λ within the mask range, and take a certain percentile Pe2 of M as the mapping threshold λ according to the brightness and darkness of the above image. Design the gray mapping function V p : e :

[0066]

[0067] where V e is the pixel value after mapping processing, V0 is the pixel value before mapping processing, and λ is the threshold.

[0068] In another specific embodiment, in step S121, before setting a mask for the initial two-dimensional sonar image, in order to improve the reliability of the initial two-dimensional sonar image, it is also necessary to perform image denoising on the initial two-dimensional sonar image.

[0069] Specifically, image denoising is performed through Kalman filtering. Kalman filtering is an existing technology and will not be elaborated here.

[0070] Furthermore, in order to perform skeletonization processing on the initial two-dimensional sonar image, first, the pixels of the two-dimensional enhanced sonar image are binarized to obtain a binarized image; then, skeleton line calculation processing is performed on the binarized image to obtain a two-dimensional skeletonized sonar image.

[0071] In this embodiment, through the skeletonization operation, not only the linear features of the two-dimensional enhanced sonar image are highlighted, which helps to extract the linear features of the target from complex images, and the extraction of such linear features can make the algorithm easier to recognize and understand the main structures and features in the image, thereby improving the accuracy of target detection and analysis; it also helps to reduce redundant information in the image, thereby simplifying subsequent processing steps. By extracting the backbone of the target, the complexity of processing can be reduced and the efficiency can be improved.

[0072] Further, in order to perform Canny edge feature extraction on the initial two-dimensional sonar image, specifically, the Canny operator is used to perform edge extraction processing on the two-dimensional skeletonized sonar image, highlighting the edge features of the two-dimensional enhanced sonar image to obtain a two-dimensional edge sonar image.

[0073] In this embodiment, detecting edges in the image through the Canny edge detection algorithm is not only crucial for enhancing the contours of underwater objects and terrain. By accurately identifying and marking these edges, the shape and position of the target can be more clearly defined; the Canny operator can also effectively reduce noise interference in the image, ensuring that the processed information is more reliable and consistent.

[0074] Further, in order to perform Hough line removal processing on the initial two-dimensional sonar image, as Figure 3 shown, Figure 3 FIG. is a schematic flowchart of an embodiment of the present invention for performing Hough line removal processing on the initial two-dimensional sonar image, including:

[0075] Step S221: Perform probabilistic Hough line processing on the two-dimensional edge sonar image to obtain a set of line segments of the two-dimensional edge sonar image;

[0076] Step S222: Set a first line segment length threshold, a second line segment length threshold, and a feature line segment;

[0077] Step S223: Perform removal processing on the set of line segments according to the first line segment length threshold, the second line segment length threshold, the feature line segment, and a preset line segment removal relationship to obtain a filtered two-dimensional sonar image.

[0078] In this embodiment, first, perform probabilistic Hough line processing on the two-dimensional edge sonar image to obtain a set of line segments of the two-dimensional edge sonar image; then, set a first line segment length threshold, a second line segment length threshold, and a feature line segment; finally, perform removal processing on the set of line segments according to the first line segment length threshold, the second line segment length threshold, the feature line segment, and a preset line segment removal relationship to obtain a filtered two-dimensional sonar image.

[0079] Among them, the preset line segment removal relationship is:

[0080]

[0081] Among them, Length L is the length of line segment L, Q1 is the first line segment length threshold, Q2 is the second line segment length threshold, & represents the AND relationship, L1, L2, L3 are the feature line segments of line segment L, means there exists, and l is any one of L1, L2, L3.

[0082] In this embodiment, by performing Hough line rejection processing on the initial two-dimensional sonar image, it has significant computational efficiency when detecting linear structures in an underwater planar environment. Since it randomly selects points on the image to estimate the line parameters without detecting all points in the image, this greatly reduces the computational complexity and makes the detection process faster. Additionally, since it does not need to traverse the entire image space, this means it can maintain high efficiency when processing large images. This is particularly important for processing challenging large underwater image data.

[0083] In a specific embodiment, based on the set H containing a series of line segments obtained by Hough line rejection processing, a method combining a length threshold with three key feature lines is designed to screen the pixel information in the processed image and filter out the circular clutter in the image. Two length thresholds are defined, namely Q1 and Q2, both of which are a certain percentile of the lengths of all line segments in the set H, and Q1 < Q2. Three key line segments are also defined as the features of each line segment element in the image. They are respectively:

[0084] The line connecting the left endpoint of line segment L and the origin O of the sonar image coordinate system;

[0085] The line connecting the right endpoint of line segment L and the origin O of the sonar image coordinate system;

[0086] The line connecting the midpoint of line segment L and the origin O of the sonar image coordinate system.

[0087] When the length of line segment L in the image is less than Q1, or when L is less than Q2 and any one of the three feature line segments L1, L2, and L3 belonging to line segment L is perpendicular to L, it is considered that L is a circular clutter line segment and should be removed.

[0088] In summary, by sequentially performing double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line rejection processing on the initial two-dimensional sonar image, it is possible to efficiently highlight the key features of the target, enhance the contour, reduce noise interference, and quickly detect linear structures in an underwater environment, thereby providing a powerful tool in underwater image processing and analysis and contributing to more accurate and reliable target detection and environmental analysis.

[0089] As a preferred embodiment, in step S103, after obtaining the filtered two-dimensional sonar image, it is also necessary to perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data. Specifically, the filtered two-dimensional sonar image is sequentially subjected to spatial transformation, filtering processing, downsampling processing, and point cloud fusion processing to obtain sonar three-dimensional point cloud data.

[0090] In a specific embodiment, a point cloud preprocessing process combining filtering and downsampling based on the Open3D method is designed. A point cloud preprocessing framework combining Radius filtering and Voxel downsampling is constructed to process the initial point cloud data. While denoising, it reduces the complex data volume and alleviates the computational burden.

[0091] Design a three-dimensional point cloud fusion algorithm, an algorithm that fuses point clouds obtained from multiple batches of images. During the fusion process, the operation of combining Radius filtering and Voxel downsampling is performed again to reduce the redundancy and noise interference of the point cloud, and obtain the final three-dimensional point cloud result to complete the three-dimensional reconstruction task.

[0092] In this embodiment, by processing the two-dimensional sonar images in batches to extract the key feature points in the images and converting these feature points into three-dimensional point clouds, the three-dimensional reconstruction task of underwater targets is realized, thus successfully solving the problems of comprehensive detection, highlighting, and accurate extraction of image features in sonar image target detection and three-dimensional reconstruction.

[0093] Obviously, the above method can be applied to fields such as underwater detection, ocean science research, and resource exploration, providing an efficient, accurate, and feasible solution for target reconstruction in the underwater environment.

[0094] Through the above method, by sequentially performing double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line elimination processing on the initial two-dimensional sonar image. Among them, the double-threshold segmentation enhancement processing can highlight the essential features of the initial two-dimensional sonar image to exclude the interference of system factors; the skeletonization processing helps to extract linear features and improve the accuracy of subsequent data analysis; the Canny edge feature extraction can better ensure the integrity of the edge features of the image; the Hough line elimination can specifically remove the irrelevant line segments in the image, thereby reducing the subsequent computational complexity. Therefore, the above image processing steps can effectively ensure the integrity of the two-dimensional sonar image features, and further ensure the reliability of the sonar three-dimensional point cloud data.

[0095] To solve the above problems, the present invention also provides a sonar image optimization system, as Figure 4 shown, Figure 4 is a structural block diagram of an embodiment of the sonar image optimization system provided by the present invention. The sonar image optimization system 400 includes:

[0096] An initial two-dimensional sonar image acquisition module 401, used to acquire an initial two-dimensional sonar image;

[0097] A filtered two-dimensional sonar image determination module 402, used to sequentially perform double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line elimination processing on the initial two-dimensional sonar image to determine the filtered two-dimensional sonar image;

[0098] The sonar three-dimensional point cloud data acquisition module 403 is used to perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

[0099] The present invention also correspondingly provides a sonar image optimization device, as Figure 5 shown Figure 5 is a structural block diagram of an embodiment of the sonar image optimization device provided by the present invention. The sonar image optimization device 500 may be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The sonar image optimization device 500 includes a processor 501 and a memory 502. Among them, a sonar image optimization program 503 is stored on the memory 502.

[0100] In some embodiments, the memory 502 may be an internal storage unit of a computer device, such as a hard disk or memory of a computer device. In other embodiments, the memory 502 may also be an external storage device of a computer device, such as a plug-in hard disk equipped on a computer device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 502 may also include both an internal storage unit and an external storage device of a computer device. The memory 502 is used to store application software installed on the computer device and various types of data, such as program codes installed on the computer device. The memory 502 may also be used to temporarily store data that has been output or will be output. In one embodiment, the sonar image optimization program 503 can be executed by the processor 501, so as to implement the sonar image optimization methods, systems, devices, and storage media of various embodiments of the present invention.

[0101] In some embodiments, the processor 501 may be a Central Processing Unit (CPU), a microprocessor, or other data processing chips, and is used to run program codes stored in the memory 502 or process data, such as executing the sonar image optimization program, etc.

[0102] This embodiment also provides a computer-readable storage medium, on which a sonar image optimization program is stored. When the program is executed by a processor, it implements the sonar image optimization methods, systems, devices, and storage media as described above.

[0103] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Any reference to a memory, storage, database, or other medium used in the embodiments provided in this application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0104] As described above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A sonar image optimization method, characterized in that Including: Obtain an initial two-dimensional sonar image; Successively perform double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image to determine a filtered two-dimensional sonar image; Perform three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data; Among them, successively perform double-threshold segmentation enhancement processing, skeletonization processing, and Canny edge feature extraction on the initial two-dimensional sonar image to obtain a two-dimensional edge sonar image; Subsequently, perform probabilistic Hough line processing on the two-dimensional edge sonar image to obtain a line segment set of the two-dimensional edge sonar image; Set a first line segment length threshold, a second line segment length threshold, and a feature line segment; Perform removal processing on the line segment set according to the first line segment length threshold, the second line segment length threshold, the feature line segment, and a preset line segment removal relationship to obtain the filtered two-dimensional sonar image; The preset line segment removal relationship is: Wherein, the is the length of line segment L, Q1 is the first line segment length threshold, and Q2 is the second line segment length threshold. represents the AND relationship. is the characteristic line segment of line segment L. represents existence, l is any one of, L1 is the line connecting the left end point of line segment L and the origin O of the sonar image coordinate system, L2 is the line connecting the right end point of line segment L and the origin O of the sonar image coordinate system, and L3 is the line connecting the midpoint of line segment L and the origin O of the sonar image coordinate system.

2. The sonar image optimization method according to claim 1, characterized in that The performing double-threshold segmentation enhancement processing on the initial two-dimensional sonar image includes: Set a mask for the initial two-dimensional sonar image, and obtain effective image pixels of the initial two-dimensional sonar image based on the mask; Set a brightness threshold, and determine a bright image set in the initial two-dimensional sonar image corresponding to the effective image pixels according to the brightness threshold; Set a mapping threshold in the bright image set, determine a gray mapping function based on the mapping threshold, and perform mapping processing on the effective image pixels according to the gray mapping function to obtain a two-dimensional enhanced sonar image.

3. The sonar image optimization method according to claim 2, wherein The performing skeletonization processing on the initial two-dimensional sonar image includes: Perform binarization processing on the pixels of the two-dimensional enhanced sonar image to obtain a binarized image; Perform skeleton line calculation processing on the binarized image to obtain a two-dimensional skeletonized sonar image.

4. The sonar image optimization method according to claim 3, wherein, The performing Canny edge feature extraction on the initial two-dimensional sonar image includes: Perform edge extraction processing on the two-dimensional skeletonized sonar image through a Canny operator to highlight the edge features of the two-dimensional enhanced sonar image and obtain the two-dimensional edge sonar image.

5. The sonar image optimization method according to claim 1, characterized in that, The performing three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data includes: Successively perform spatial transformation, filtering processing, downsampling processing, and point cloud fusion processing on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data.

6. A sonar image optimization system, characterized in that, Including: An initial two-dimensional sonar image acquisition module for acquiring an initial two-dimensional sonar image; A filtered two-dimensional sonar image determination module for successively performing double-threshold segmentation enhancement processing, skeletonization processing, Canny edge feature extraction, and Hough line removal processing on the initial two-dimensional sonar image to determine a filtered two-dimensional sonar image; A sonar three-dimensional point cloud data acquisition module for performing three-dimensional reconstruction on the filtered two-dimensional sonar image to obtain sonar three-dimensional point cloud data; Among them, successively perform double-threshold segmentation enhancement processing, skeletonization processing, and Canny edge feature extraction on the initial two-dimensional sonar image to obtain a two-dimensional edge sonar image; Subsequently, perform probabilistic Hough line processing on the two-dimensional edge sonar image to obtain a line segment set of the two-dimensional edge sonar image; Set a first line segment length threshold, a second line segment length threshold, and a feature line segment; Perform an elimination process on the line segment set according to the first line segment length threshold, the second line segment length threshold, the feature line segment, and a preset line segment elimination relationship to obtain the filtered two-dimensional sonar image; The preset line segment elimination relationship is: Among them, the is the length of line segment L, Q1 is the first line segment length threshold, and Q2 is the second line segment length threshold. represents the AND relationship. is the characteristic line segment of line segment L. represents existence, and l is any one of, L1 is the line connecting the left end point of line segment L and the origin O of the sonar image coordinate system, L2 is the line connecting the right end point of line segment L and the origin O of the sonar image coordinate system, and L3 is the line connecting the midpoint of line segment L and the origin O of the sonar image coordinate system.

7. A sonar image optimization device, characterized in that, Comprising a memory and a processor, wherein, The memory is used for storing a program; The processor is coupled to the memory and is used for executing the program stored in the memory to implement the steps in the sonar image optimization method described in any one of claims 1 to 5 above.

8. A computer-readable storage medium, characterized in that, For storing computer-readable programs or instructions, the programs or instructions can implement the steps in the sonar image optimization method described in any one of claims 1 to 5 above when executed by a processor.

Images