Side-scan sonar image submarine cable accurate segmentation method

Through the method based on curved wave transformation and Hough transformation, the color loss and edge blur of cable segmentation in side-sweep sonar images are solved, and the precise segmentation effect is achieved without a large amount of data and prior knowledge, which is suitable for different sizes and resolutions Scan the sonar image of the seabed side.

CN119068013AActive Publication Date: 2024-12-03NORTHWESTERN POLYTECHNICAL UNIV

Patent Information

Application Number
CN202411253616.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-09
Publication Date
2024-12-03
Estimated Expiration
2044-09-09

AI Technical Summary

Technical Problem

The existing technology has problems such as color loss, low contrast, and blurred edges in cable segmentation tasks in side-scan sonar images, which leads to difficulty in target segmentation. Deep learning methods require a large amount of data and high cost training, while traditional methods are specific. The image size and resolution are limited.

Method used

Image filtering is performed by using a cable feature automatic extraction filter based on curved wave transform, combined with Hough transformation and two-dimensional average constant alarm algorithm for noise cancellation and binarization, and convolution operation is performed using linear structural units to restore lines in a morphological manner. Cable objectives to achieve precise segmentation without prior knowledge and large amounts of data training.

Benefits of technology

The precise segmentation of cables in the subsea side-swept sonar image is achieved, reducing the image clarity and size requirements, and is suitable for segmenting images with different backgrounds and resolutions without setting thresholds and manual intervention.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068013A_ABST
    Figure CN119068013A_ABST
Patent Text Reader

Abstract

The invention relates to a side-scan sonar image submarine cable accurate segmentation method, which comprises the steps of inputting an image I1, processing the image I1 by using a curvelet transform-based cable feature automatic extraction filter, and extracting information related to a cable to obtain a feature image I2; multiplying the image I1 by the feature image I2 pixel by pixel to obtain a feature image I3; finding out not less than three straight line segments in the feature image I3 by using Hough transform, recording an angle alpha of the longest line segment in the image, constructing a linear structure unit in the same direction as the cable in the image, and performing noise elimination and binarization processing on the feature image I3 to obtain a binary image I4; performing cable form restoration on the binary image I4 to obtain a cable form restoration image I5, and only reserving the maximum connected area in the image as a submarine side-scan sonar image cable segmentation result; the problems that in side-scan sonar image segmentation, a large amount of data is needed for training, the requirement for image definition is high, target color information needs to be obtained, and the requirement for the size of an image is high are solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the technical field of side-scan sonar image processing, and in particular relates to a method for accurately segmenting submarine cables using side-scan sonar images. Background Art

[0002] In modern marine affairs, submarine cables are responsible for important tasks such as resource transportation and information transmission. Side-scan sonar is the most important tool for humans to perceive the seabed. Identifying and segmenting cables on side-scan sonar images is the basis for tracking, maintaining and repairing them. Side-scan sonar images generally have problems such as color loss, low contrast, and blurred edges, which bring difficulties to the target segmentation task.

[0003] The segmentation techniques proposed so far can be divided into two categories: deep learning-based image segmentation methods and traditional image segmentation methods. Deep learning-based methods, such as: Bharti, V., D. Lane and S. Wang. "Learning to detect subsea pipelines with deep segmentation network and self-supervision," Global Oceans 2020: Singapore-US Gulf Coast, pp. 1-7, 2020. A deep neural network was designed to segment cable targets from optical images obtained from highly turbid seabed scenes, which are similar to sonar images; Du X, Sun Y, Song Y, Dong L, Zhao X. "Revealing the Potential of Deep Learning for Detecting Submarine Pipelines in Side-Scan Sonar Images: An Investigation of Pre-Training Datasets," Measurement Science and Technology, vol. 15, no. 19, p. 4873, 2023. GoogleNet was used to automatically detect and identify pipelines or cables from side-scan sonar images. However, deep learning-based methods require a large amount of data to fully train the network, and require huge costs for image acquisition and manual annotation. In addition, deep learning-based methods usually require the input image to be of a fixed size. Even after multi-scale training, the model can only be applied to images of a specific size as input, which limits the scope of application of the algorithm to a certain extent.

[0004] Traditional image segmentation methods are also used in sonar image cable segmentation tasks. Paredes GM, Taveira-Pinto F. "An experimental technique to track mooring cables in smallscale models using image processing," Ocean Engineering, vol. 111, pp. 439-448, 2016. A method for segmenting cables in an image is proposed, using top-hat transformation to eliminate small noise points, and using Canny edge detector and morphological dilation to segment cable targets; "An underwater sonar image target segmentation method based on an adaptive cultural algorithm" (application number: CN201811086885.6, publication date: 2019-03-08) proposes an adaptive cultural algorithm to segment objects in side-scan sonar images; "A submarine pipeline detection and positioning method and system based on forward-looking sonar" (application number: CN202110891976.2, publication date: 2023-07-18) performs dynamic threshold detection, binarization processing, pipeline segment extraction and other operations on multi-beam domain data to achieve detection and segmentation and positioning of pipeline targets in sonar images. Traditional image segmentation methods can be divided into clustering methods based on color differences and segmentation methods based on target edge feature extraction. The former, when faced with color-deficient side-scan sonar images, only relies on the size of pixel values ​​for clustering segmentation and is unable to distinguish between cable targets and interfering objects. Although the latter shows good effects in optical images, it usually cannot achieve satisfactory results when applied to seabed sonar images with low resolution and blurred edges. Summary of the invention

[0005] In order to overcome the above problems, the purpose of the present application is to propose a method for accurately segmenting submarine cables in side-scan sonar images, which can accurately segment the cables in side-scan sonar images.

[0006] In order to achieve the above purpose, the technical solution adopted in this application is as follows:

[0007] A method for accurately segmenting submarine cables in side-scan sonar images, comprising:

[0008] Step 1: Input a seabed side-scan sonar image I1 containing cables, use a cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing cables, extract information related to the cables in the image, and obtain a feature image I2;

[0009] Step 2: Multiply the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained in step 1 pixel by pixel to strengthen the cable feature while weakening the interference information in the background, thereby obtaining the feature image I3;

[0010] Step 3: For the feature image I3 obtained in step 2, use Hough transform to find no less than three straight line segments, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image;

[0011] Step 4: Using the direction angle α of the cable in the image obtained in step 3, construct a linear structural unit in the same direction as the cable in the image;

[0012] Step 5: De-noise and binarize the feature image I3 obtained in step 2 to obtain a binary image I4;

[0013] Step 6: Perform cable shape restoration on the binary image I4 obtained in step 5, perform a convolution operation on the binary image I4 using the linear structure unit obtained in step 4, connect the pixel regions of the cable targets in the image, and obtain a cable shape restoration image I5;

[0014] Step 7: For the cable shape restoration image I5 obtained in step 6, only the largest connected area in the image is retained as the cable segmentation result of the seabed side-scan sonar image.

[0015] In step 1, the cable feature automatic extraction filter based on curvelet transform is used to filter the seabed side-scan sonar image I1 containing the cable, specifically:

[0016] The side-scan sonar image I1 containing cables is subjected to curvelet 7-level decomposition, and the numbers of directional divisions of levels 1-7 are 1, 16, 32, 32, 64, 64, and 128, respectively. After curvelet decomposition, two sets of coefficients symmetrically located at each level are used to reconstruct an image. The coefficients of levels 1-2 are directly filtered out, and the variance selection method is used for the coefficients of levels 3-7. The coefficients of the six areas with the largest variance are selected in each level to be retained, and the other coefficients are filtered out.

[0017] In step 5, a two-dimensional average constant false alarm algorithm is first used to filter out noise points from the feature image I3 obtained in step 2, and then the denoised image is binarized to obtain a binary image I4.

[0018] In step 6, the linear structure unit obtained in step 4 is used to perform a convolution operation on the binary image I4, and the cable targets in the image are connected in pixel areas to obtain a cable morphology restoration image I5. Specifically, the linear structure unit obtained in step 4 is used as a mask to perform a convolution operation with the binary image I4, and the cables that are interrupted in the binary image I4 are reconnected to obtain the cable morphology restoration image I5.

[0019] When the two-dimensional average constant false alarm algorithm is used to filter out noise points on the feature image I3 obtained in step 2, the false alarm probability is set to 0.05. When the denoised image is binarized, the pixel values ​​of pixels with pixel values ​​greater than 0 are all set to 1.

[0020] A side scan sonar image submarine cable precise segmentation system, comprising:

[0021] Input module: input the seabed side-scan sonar image I1 containing the cable, use the cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing the cable, extract the information related to the cable in the image, and obtain the feature image I2;

[0022] The first processing module: multiplies the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained by the input module pixel by pixel to strengthen the cable feature while weakening the interference information in the background, thereby obtaining the feature image I3;

[0023] The second processing module: for the feature image I3 obtained by the first processing module, use Hough transform to find no less than three straight line segments, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image;

[0024] The third processing module: using the direction angle α of the cable in the image obtained in the second processing module, constructing a linear structure unit in the same direction as the cable in the image;

[0025] The fourth processing module: performs noise elimination and binarization processing on the feature image I3 obtained by the first processing module to obtain a binary image I4;

[0026] Cable shape restoration module: performs cable shape restoration on the binary image I4 obtained by the fourth processing module, performs convolution operation on the binary image I4 using the linear structure unit obtained by the third processing module, connects the pixel areas of the cable targets in the image, and obtains the cable shape restoration image I5;

[0027] Output module: For the cable shape restoration image I5 obtained by the cable shape restoration module, only the largest connected area in the image is retained as the cable segmentation result of the seabed side-scan sonar image.

[0028] An electronic device, comprising:

[0029] at least one processor;

[0030] and, a memory communicatively coupled to the at least one processor;

[0031] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method for accurately segmenting submarine cables in side-scan sonar images.

[0032] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, a method for accurately segmenting submarine cables in side-scan sonar images is implemented.

[0033] Compared with the prior art, the beneficial effects of this application are:

[0034] 1. This application designs a line feature extraction filter based on curvelet transform according to the straight line characteristics of artificially laid submarine cables in side-scan sonar images, which can automatically extract cable information in side-scan sonar images while weakening interference information in the background.

[0035] 2. This application designs an image binarization method that does not require a threshold to be set in advance. Before the binarization operation, a two-dimensional average constant false alarm algorithm is used to eliminate noise points in the feature image, and the pixel values ​​of the pixels in the denoised feature image whose pixel values ​​are greater than 0 are set to 1. Since there is no need to set a threshold, there is no need to extract prior information from a large number of acquired images, and the influence of the rationality of the threshold setting on the experimental results is avoided.

[0036] 3. This application designs a method for automatic restoration of line feature morphology, which can automatically restore the morphology of cables in any direction in the image without human intervention, use Hough transform to determine the angle information of the cable in the image, and then construct a linear structure unit in the same direction as the cable. Finally, the constructed linear structure unit is used to perform a convolution operation with the feature image, which can restore the morphology of non-continuous cables in the feature image and reconnect them to obtain a complete cable target.

[0037] In summary, the algorithms involved in this application are improved or combined with each other based on the characteristics of artificially laid submarine cables in side-scan sonar images. The overall algorithm does not require prior knowledge as a guide, does not require a large amount of data for training, has low requirements for image clarity, does not need to obtain target color information, and has no special requirements for the size of the image. It is suitable for segmenting cable targets in seabed side-scan sonar images. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flow chart of the method of this application;

[0039] Figure 2 This is a schematic diagram of the image of this application being divided by the curvelet 7-level decomposition coefficients;

[0040] Figure 3 This is the seabed side-scan sonar image of the embodiment of the present application;

[0041] Figure 4 is the normalized variance of the curvelet coefficients in each direction of levels 3 to 7 of the embodiment of the present application;

[0042] Figure 5 The image reconstructed by the 4th level curvelet coefficients of the embodiment of the present application;

[0043] Figure 6 It is a feature map of the image of the embodiment of the present application after the image is subjected to the line feature automatic extraction filter based on the curvelet transform;

[0044] Figure 7 It is a feature image obtained by fusing the feature map after the line feature automatic extraction filter based on curvelet transform in the embodiment of the present application with the original image;

[0045] Figure 8 The image is subjected to denoising and binarization processing in the embodiment of the present application;

[0046] Fig. 9 In the embodiment of the present application, Hough transform is used to construct a linear structural unit in the same direction as the cable;

[0047] Fig.10 This is the morphological restoration result of the embodiment of the present application;

[0048] Fig.11 This is the cable target segmentation result of the embodiment of the present application;

[0049] Fig.12 This is a comparison chart of the side scan sonar image and cable target segmentation results in an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present application more clear, the present application is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

[0051] like Figure 1 As shown, a method for accurately segmenting submarine cables in side scan sonar images comprises:

[0052] A method for accurately segmenting submarine cables in side-scan sonar images, comprising:

[0053] Step 1: Input a seabed side-scan sonar image I1 containing cables, use a cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing cables, extract information related to the cables in the image, and obtain a feature image I2;

[0054] Step 2: Multiply the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained in step 1 pixel by pixel to strengthen the cable feature while weakening the interference information in the background, thereby obtaining the feature image I3;

[0055] Step 3: For the feature image I3 obtained in step 2, use Hough transform to find no less than three straight line segments, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image;

[0056] Step 4: Using the direction angle α of the cable in the image obtained in step 3, construct a linear structural unit in the same direction as the cable in the image;

[0057] Step 5: De-noise and binarize the feature image I3 obtained in step 2 to obtain a binary image I4;

[0058] Step 6: Perform cable shape restoration on the binary image I4 obtained in step 5, perform a convolution operation on the binary image I4 using the linear structure unit obtained in step 4, connect the pixel regions of the cable targets in the image, and obtain a cable shape restoration image I5;

[0059] Step 7: For the cable shape restoration image I5 obtained in step 6, only the largest connected area in the image is retained as the cable segmentation result of the seabed side-scan sonar image.

[0060] In step 6, the linear structure unit obtained in step 4 is used to perform a convolution operation on the binary image I4, and the cable targets in the image are connected in pixel areas to obtain a cable morphology restoration image I5. Specifically, the linear structure unit obtained in step 4 is used as a mask to perform a convolution operation with the binary image I4, and the cables that are interrupted in the binary image I4 are reconnected to obtain the cable morphology restoration image I5.

[0061] Design of the automatic line feature extraction filter based on curvelet transform in step 1:

[0062] The image is decomposed into 7 levels by curvelet. The number of directional divisions of levels 1 to 7 are 1, 16, 32, 32, 64, 64, and 128 respectively. The coefficients after curvelet decomposition are divided as follows: Figure 2 As shown in the figure, it should be noted that after the image is decomposed by curvelet, two sets of coefficients symmetrically located at each level reconstruct an image; the experiment found that after the seabed side-scan sonar image containing cables is decomposed by curvelet at level 7, the low-frequency coefficients at levels 1-2 contain a lot of background information, while the cable information is concentrated in the high-frequency coefficients at levels 3-7 in specific directions; the coefficients at levels 1-2 are directly filtered out, and the variance selection method is used for the coefficients at levels 3-7. The coefficients in the six areas with the largest variance are selected at each level to be retained, and the other coefficients are filtered out.

[0063] In step 5, the feature image I3 obtained in step 2 is firstly filtered out using the two-dimensional average constant false alarm algorithm, and then the denoised image is binarized. In the two-dimensional average constant false alarm algorithm, the false alarm probability is set to 0.05; in the image binarization method, since the image has been subjected to denoising, there is no need to set an additional binarization threshold, and the values ​​of the pixels with pixel values ​​greater than 0 are all set to 1.

[0064] In step 6, the binary image I4 is convolved with the linear structure unit obtained in step 4, and the cable targets in the image are connected in pixel areas to obtain the cable morphology restoration image I5. Specifically, using the direction angle of the cable in the image obtained in step 3, a linear structure unit with a length of 40 pixels is constructed in step 4, and the binary image I4 is convolved with the linear structure unit to connect the pixel points representing the cable.

[0065] Example

[0066] This example is about Figure 3 Perform pipeline target segmentation.

[0067] (1) First, the image is filtered using a line feature automatic extraction filter based on curvelet transform. The variance of the curvelet coefficients in each direction of level 3-7 is as follows: Figure 4 As shown in the figure, the variance value is normalized at each scale. Taking the coefficients of level 4 as an example, the three images reconstructed by the six groups of coefficients selected by the variance selection method are as follows: Figure 5 As shown; the feature map of the image after curvelet filtering is as follows Figure 6 shown.

[0068] (2) Multiply the original image and the filtered feature map pixel by pixel to obtain the fused feature map, such as Figure 7 shown.

[0069] (3) Using the two-dimensional average constant false alarm algorithm Figure 7 Noise elimination is performed, and the false alarm probability is set to 0.05; then a binarization operation is performed, and the pixel values ​​of pixels with pixel values ​​greater than 0 are set to 1, and the following is obtained: Figure 8 The binary image shown.

[0070] (4) Yes Figure 7 Apply Hough transform to detect the three straight line segments and record the angle α of the longest line segment in the image as the direction angle of the cable in the image. Based on this, the linear structure unit is constructed. This process is as follows: Fig. 9 shown.

[0071] (5) Morphological restoration of cables: Fig. 9 The linear structural unit constructed in Figure 8The binary image after noise reduction is convolved to connect the pixel areas of the cable targets that may be interrupted to ensure the integrity of the final segmentation result. The morphological restoration result is as follows: Fig.10 shown.

[0072] (6) Only retain Fig.10 The largest connected area in is taken as the cable target segmentation result, such as Fig.11 shown.

[0073] like Fig.12 As shown, the left side figures (a1), (a2), (a3), (a4), (a5) and (a6) are side scan sonar images, and the right side figures (b1), (b2), (b3), (b4), (b5) and (b6) are corresponding cable target segmentation result figures. This application performs target segmentation on 6 seabed side scan sonar images containing cables and with many interference targets. The 6 side scan sonar images are acquired under different background and distance conditions. The background contains a lot of interference information such as seabed rocks, sand textures, uneven terrain, etc., and the overall image resolution is low. The algorithms designed in this application can accurately segment the cable targets therein.

[0074] A side scan sonar image submarine cable precise segmentation system, comprising:

[0075] Input module: input the seabed side-scan sonar image I1 containing the cable, use the cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing the cable, extract the information related to the cable in the image, and obtain the feature image I2;

[0076] The first processing module: multiplies the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained by the input module pixel by pixel, so as to strengthen the cable feature while weakening the interference information in the background, and obtain the feature image I3;

[0077] The second processing module: for the feature image I3 obtained by the first processing module, use Hough transform to find no less than three straight line segments in the feature image I3, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image;

[0078] The third processing module: using the direction angle α of the cable in the image obtained in the second processing module, constructing a linear structure unit in the same direction as the cable in the image;

[0079] The fourth processing module: performs noise elimination and binarization processing on the feature image I3 obtained by the first processing module to obtain a binary image I4;

[0080] Cable shape restoration module: performs cable shape restoration on the binary image I4 obtained by the fourth processing module, performs convolution operation on the binary image I4 using the linear structure unit obtained by the third processing module, connects the pixel areas of the cable targets in the image, and obtains the cable shape restoration image I5;

[0081] Output module: For the cable shape restoration image I5 obtained by the cable shape restoration module, only the largest connected area in the image is retained and output as the cable segmentation result of the seabed side-scan sonar image.

[0082] A terminal / electronic device / server, comprising:

[0083] at least one processor;

[0084] and, a memory communicatively coupled to the at least one processor;

[0085] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method for accurately segmenting submarine cables in side-scan sonar images.

[0086] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, a method for accurately segmenting submarine cables in side-scan sonar images is implemented.

[0087] It is not difficult to find that this embodiment is a system embodiment corresponding to the above method embodiment, and this embodiment can be implemented in conjunction with the above method embodiment. The relevant technical details and technical effects mentioned in the above embodiments are still valid in this embodiment, and in order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiments.

[0088] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0089] Another embodiment of the present application relates to an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for accurately segmenting submarine cables in side-scan sonar images in the above-mentioned embodiments.

[0090] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0091] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0092] Another embodiment of the present application relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0093] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM (Read-Only Memory), RAM (Random Access Memory), disk or optical disk and other media that can store program codes.

[0094] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. A method for accurately segmenting submarine cables from side-scan sonar images, characterized in that: include: Step 1: Input a seabed side-scan sonar image I1 containing cables, use a cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing cables, extract information related to the cables in the image, and obtain a feature image I2; Step 2: Multiply the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained in step 1 pixel by pixel to strengthen the cable feature while weakening the interference information in the background, thereby obtaining the feature image I3; Step 3: For the feature image I3 obtained in step 2, use Hough transform to find no less than three straight line segments, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image; Step 4: Using the direction angle α of the cable in the image obtained in step 3, construct a linear structural unit in the same direction as the cable in the image; Step 5: De-noise and binarize the feature image I3 obtained in step 2 to obtain a binary image I4; Step 6: Perform cable shape restoration on the binary image I4 obtained in step 5, perform a convolution operation on the binary image I4 using the linear structure unit obtained in step 4, connect the pixel regions of the cable targets in the image, and obtain a cable shape restoration image I5; Step 7: For the cable shape restoration image I5 obtained in step 6, only the largest connected area in the image is retained as the cable segmentation result of the seabed side-scan sonar image.

2. The method for accurately segmenting submarine cables from side scan sonar images according to claim 1, characterized in that: In step 1, the cable feature automatic extraction filter based on curvelet transform is used to filter the seabed side-scan sonar image I1 containing the cable, specifically: The side-scan sonar image I1 containing cables is subjected to curvelet 7-level decomposition, and the numbers of directional divisions of levels 1-7 are 1, 16, 32, 32, 64, 64, and 128, respectively. After curvelet decomposition, two sets of coefficients symmetrically located at each level are used to reconstruct an image. The coefficients of levels 1-2 are directly filtered out, and the variance selection method is used for the coefficients of levels 3-7. The coefficients of the six areas with the largest variance are selected in each level to be retained, and the other coefficients are filtered out.

3. The method for accurately segmenting submarine cables from side scan sonar images according to claim 1, characterized in that: In step 5, a two-dimensional average constant false alarm algorithm is first used to filter out noise points from the feature image I3 obtained in step 2, and then the denoised image is binarized to obtain a binary image I4.

4. The method for accurately segmenting submarine cables from side scan sonar images according to claim 1, characterized in that: In step 6, the linear structure unit obtained in step 4 is used to perform a convolution operation on the binary image I4, and the cable targets in the image are connected in pixel areas to obtain a cable morphology restoration image I5. Specifically, the linear structure unit obtained in step 4 is used as a mask to perform a convolution operation with the binary image I4, and the cables that are interrupted in the binary image I4 are reconnected to obtain the cable morphology restoration image I5.

5. The method for accurately segmenting submarine cables from side scan sonar images according to claim 3 is characterized in that: When the two-dimensional average constant false alarm algorithm is used to filter out noise points on the feature image I3 obtained in step 2, the false alarm probability is set to 0.

05. When the denoised image is binarized, the pixel values ​​of pixels with pixel values ​​greater than 0 are all set to 1.

6. A side scan sonar image submarine cable accurate segmentation system, characterized in that: include: Input module: input the seabed side-scan sonar image I1 containing the cable, use the cable feature automatic extraction filter based on curvelet transform to filter the seabed side-scan sonar image I1 containing the cable, extract the information related to the cable in the image, and obtain the feature image I2; The first processing module: multiplies the seabed side-scan sonar image I1 containing the cable by the feature image I2 obtained by the input module pixel by pixel to strengthen the cable feature while weakening the interference information in the background, thereby obtaining the feature image I3; The second processing module: for the feature image I3 obtained by the first processing module, use Hough transform to find no less than three straight line segments, and record the angle α of the longest line segment in the image as the direction angle of the cable in the image; The third processing module: using the direction angle α of the cable in the image obtained in the second processing module, constructs a linear structure unit in the same direction as the cable in the image; The fourth processing module: performs noise elimination and binarization processing on the feature image I3 obtained by the first processing module to obtain a binary image I4; Cable shape restoration module: performs cable shape restoration on the binary image I4 obtained by the fourth processing module, performs convolution operation on the binary image I4 using the linear structure unit obtained by the third processing module, connects the pixel areas of the cable targets in the image, and obtains the cable shape restoration image I5; Output module: For the cable shape restoration image I5 obtained by the cable shape restoration module, only the largest connected area in the image is retained as the cable segmentation result of the seabed side-scan sonar image.

7. An electronic device, characterized in that: include: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for accurately segmenting submarine cables in side-scan sonar images according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for accurately segmenting submarine cables in side-scan sonar images according to any one of claims 1 to 5 is implemented.

Citation Information

Patent Citations

  • Underwater small target detection method and system of sonar image under complex background

    CN115170941A

  • Submarine cable detection and identification method based on side-scan sonar

    CN116523822A

  • Systems, methods, and devices for capturing images

    US20220284691A1

Cited By

  • Construction method of adaptive semantic segmentation seabed linear target detection model

    CN120635429A

  • A method for constructing an adaptive semantic segmentation model for detecting linear targets on the seabed

    CN120635429B