A side scan sonar image based seabed elevation detection method based on prior knowledge

Abstract

The present application belongs to the technical field of marine surveying and mapping, and discloses a method for detecting the sea bottom elevation of a side scan sonar image based on prior knowledge. The method is based on the left and right side original strip images measured by a side scan sonar measuring platform, and fully utilizes the gradient characteristics, continuity, symmetry and other prior information of the sea bottom boundary to automatically extract the sea bottom line in the image, so as to measure the elevation of the sonar platform relative to the sea bottom, and lay a foundation for the conversion of the slant range information into the horizontal range information and the synthesis of the topographic image. The present application is simple and easy to operate, has robustness, and has the potential for wide application.

Description

Technical Field

[0001] This invention belongs to the field of marine surveying and mapping technology, and more specifically, relates to a method for detecting seabed elevation based on side-scan sonar images using prior knowledge. Background Technology

[0002] Side-scan sonar is one of the main instruments for underwater exploration and marine surveys, boasting advantages such as low cost, high resolution, and high scanning efficiency. Side-scan sonar emits sound waves underwater towards both port and starboard sides. These waves propagate a distance through the water and are reflected back to the seabed. The echo intensity reflects seabed characteristics; recording the echo intensity sequence in chronological order yields the underwater topography of the current cross-section. As the sonar platform moves, it continuously records the backscattering intensity of the echoes, allowing for the acquisition of underwater topographic images of the scanned area from lines to surfaces. It is evident that side-scan sonar does not store phase information, does not perform coherent positioning, and cannot directly acquire elevation information. Typically, the seabed line is extracted from the side-scan sonar image data, the platform elevation is calculated, and the target slant range is obtained based on the recorded echo time sequence. The horizontal distance is calculated from the measured elevation, thus acquiring the underwater topographic image. Therefore, the accuracy of the underwater elevation measurement at each measurement position of the side-scan sonar directly determines the positional accuracy of each point on each cross-section of the topographic image, thereby determining the accuracy of the topographic image.

[0003] The conventional method for side-scan sonar seabed elevation measurement involves manually or by setting intensity thresholds to extract the seabed line from side-scan sonar images and then calculating the seabed elevation. However, this method is easily affected by sea surface echoes or echoes from stray targets in the water, making it difficult to achieve accurate and reliable automated detection results. Manual marking methods also significantly reduce measurement efficiency and increase labor costs. Summary of the Invention

[0004] This invention provides a method for detecting seabed elevation using side-scan sonar images based on prior knowledge, thereby solving the problems of low measurement efficiency and low accuracy of seabed elevation using side-scan sonar in the prior art.

[0005] This invention provides a method for detecting seabed elevation based on prior knowledge using side-scan sonar images, comprising the following steps:

[0006] Step 1: Obtain the port side stripe image and the starboard side stripe image, and convert them into grayscale images to obtain the port side grayscale matrix IL and the starboard side grayscale matrix IR;

[0007] Step 2: Based on a preset threshold, classify the port grayscale matrix IL and the starboard grayscale matrix IR respectively to obtain the port classification matrix CL and the starboard classification matrix CR;

[0008] Step 3: Match each section in the port side classification matrix CL based on the port side template to obtain the port side correlation matrix XL; match each section in the starboard side classification matrix CR based on the starboard side template to obtain the starboard side correlation matrix XR.

[0009] Step 4: Obtain the port side continuity matrix QL based on the port side correlation matrix XL, and obtain the starboard side continuity matrix QR based on the starboard side correlation matrix XR;

[0010] Step 5: Obtain the port side symmetry matrix SL based on the port side correlation matrix XL, and obtain the starboard side symmetry matrix SR based on the starboard side correlation matrix XR;

[0011] Step 6: Combine the port side correlation matrix XL, the port side continuity matrix QL, and the port side symmetry matrix SL, and weight them to obtain the port side confidence matrix BL; combine the starboard side correlation matrix XR, the starboard side continuity matrix QR, and the starboard side symmetry matrix SR, and weight them to obtain the starboard side confidence matrix BR.

[0012] Step 7: In the port side confidence matrix BL, for each section, mark the point with the largest confidence value as the seabed boundary point of that port side section, and use the curve connecting the seabed boundary points of all port side sections as the port side seabed line; In the starboard side confidence matrix BR, for each section, mark the point with the largest confidence value as the seabed boundary point of that starboard side section, and use the curve connecting the seabed boundary points of all starboard side sections as the starboard side seabed line;

[0013] Step 8: Calculate the seabed elevation based on the starboard seabed line and the port seabed line.

[0014] Preferably, in step 1, the dimensions of the port grayscale matrix IL and the starboard grayscale matrix IR are both M×N; where M is the number of points in a cross-section and N is the number of cross-sections;

[0015] In step 2, pixels with gray values ​​greater than or equal to the threshold are classified as predicted targets with a classification value of 1; pixels with gray values ​​less than the threshold are classified as predicted background with a classification value of 0.

[0016] The port side classification matrix CL is represented as follows: The starboard classification matrix CR is represented as follows:

[0017] Where q is the threshold; i∈[1,M], i represents the section point number; j∈[1,N], j represents the section number; CL(i,j) represents the classification value of the element in the i-th row and j-th column of the port side classification matrix; IL(i,j) represents the gray value of the element in the i-th row and j-th column of the port side gray matrix; CR(i,j) represents the classification value of the element in the i-th row and j-th column of the starboard side classification matrix; and IR(i,j) represents the gray value of the element in the i-th row and j-th column of the starboard side gray matrix.

[0018] Preferably, in step 3, the port side template TL is [1 1 1 1 0 0 0], and the starboard side template TR is [00 0 1 1 1 1];

[0019] The formula for calculating the port side correlation matrix XL is as follows: The formula for calculating the starboard correlation matrix XR is as follows:

[0020] Where XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port side correlation matrix, TL(k) represents the value of the k-th element in the port side template, CL(i+k,j) represents the classification value of the element in the (i+k)-th row and j-th column of the port side classification matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard side correlation matrix, TR(k) represents the value of the k-th element in the starboard side template, and CR(i+k,j) represents the classification value of the element in the (i+k)-th row and j-th column of the starboard side classification matrix.

[0021] Preferably, in step 4, taking the i-th point L(i,j) on section j of one side as the center and the corresponding i-th point L(i,j-1) on section j-1 of the same side as the center, find the maximum correlation value in the first neighborhood formed by the n points before and after; and obtain the continuity matrix of the side based on the maximum correlation value in the first neighborhood and the correlation matrix of the side.

[0022] The formula for calculating the port side continuity matrix QL is as follows:

[0023] QL(i,j)=XL(i,j)·Max(XL(in,j-1),XL(i-n+1,j-1),…,XL(i+n,j-1);

[0024] Where QL(i,j) represents the continuity evaluation value of the element in the i-th row and j-th column of the port side continuity matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port side correlation matrix, and n represents the number of points constituting the first neighborhood.

[0025] The formula for calculating the starboard continuity matrix QR is:

[0026] QR(i,j)=XR(i,j)·Max(XR(in,j-1),XR(i-n+1,j-1),…,XR(i+n,j-1));

[0027] Where QR(i,j) represents the continuity assessment value of the element in the i-th row and j-th column of the starboard continuity matrix, and XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard correlation matrix.

[0028] Preferably, in step 5, taking the i-th point L(i,j) on the j-th section of one side as the center, i.e., the i-th point R(i,j) on the j-th section of the other side, find the maximum correlation value in the second neighborhood formed by the m points before and after; and obtain the symmetry matrix of the side based on the maximum correlation value in the second neighborhood and the correlation matrix of the side.

[0029] The formula for calculating the port side symmetry matrix SL is as follows:

[0030] SL(i,j)=XL(i,j)·Max(XR(im,j),XR(i-m+1,j),…,XR(u+m,j));

[0031] The formula for calculating the starboard symmetry matrix SR is as follows:

[0032] SR(i,j)=XR(i,j)·Max(XL(im,j),XL(i-m+1,j),…,XL(i+m,j));

[0033] Where SL(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the port symmetry matrix, SR(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the starboard symmetry matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port correlation matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard correlation matrix, and m represents the number of points constituting the second neighborhood.

[0034] Preferably, in step 6, the formula for calculating the port confidence matrix BL is: BL=(1-k1-k2)XL+k1SL+k2QL; the formula for calculating the starboard confidence matrix BR is: BR=(1-k1-k2)XR+k1SR+k2QR;

[0035] Where k1 represents the weighting factor for symmetry evaluation, and k2 represents the weighting factor for continuity evaluation.

[0036] Preferably, in step 8, the seabed elevation of the j-th section on the port side is expressed as: The seabed elevation of the j-th section on the starboard side is expressed as:

[0037] Where HL(j) represents the seabed elevation of the j-th section on the port side, HR(j) represents the seabed elevation of the j-th section on the starboard side, VL(j) represents the pixel distance between the seabed boundary point of the j-th section on the port side and the starting point of the sound source, VR(j) represents the pixel distance between the seabed boundary point of the j-th section on the starboard side and the starting point of the sound source, c represents the underwater sound speed, ΔT represents the sampling time of the side-scan sonar image data points, and f s This indicates the sampling frequency of the side-scan sonar image data points.

[0038] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0039] This invention is based on the original port and starboard strip images obtained by a side-scan sonar measurement platform. It makes full use of prior information such as the gradient features, continuity, and symmetry of the seabed boundary to automatically extract the seabed line from the image, thereby measuring the elevation of the sonar platform relative to the seabed. This lays the foundation for the next step of converting slant range information into horizontal range information and synthesizing terrain images. This invention is simple to operate, robust, and has the potential for wide application. Attached Figure Description

[0040] Figure 1 This is a side-scan sonar source strip image;

[0041] Figure 2 This is a flowchart of a method for detecting seabed elevation based on side-scan sonar images using prior knowledge, provided as an embodiment of the present invention. Detailed Implementation

[0042] The side-scan sonar emits two beams of sound waves towards the seabed. The beams are narrow in the direction of the sonar platform's advance, but wide in the transverse direction perpendicular to the advance direction, covering a transverse section. After the emitted sound waves encounter a seabed target and reflect, the side-scan sonar platform records the intensity of the reflected echoes sequentially over time, thus obtaining a one-dimensional intensity sequence for each of the two sections. During navigation, the side-scan sonar platform continuously acquires one-dimensional intensity sequences of consecutive sections on both sides of the port and starboard sides, and stacks these sequences to form two two-dimensional side-scan sonar topographic strip images covering the port and starboard side sections.

[0043] Figure 1This is a side-scan sonar source strip image. The image is directly stitched together from the port and starboard side strip images; the left half is the port side strip image, and the right half is the starboard side strip image. The centerline of the image indicates the source emission location; the darker areas on either side indicate strong echoes, usually from the seabed; the darker areas between the source and the seabed indicate no target echo received, usually from water; irregular areas of strong echoes between the source and the seabed usually indicate surface echoes or reverberation / clutter interference, which greatly affect the accurate extraction of the seabed line. The boundary between water, clutter, and the seabed is the seabed line.

[0044] To meet the requirements of automatic, accurate, and reliable extraction of seabed elevation in complex environments and improve the mapping accuracy and quality of side-scan sonar topographic images, this invention, based on the features and prior knowledge of the seabed line in side-scan sonar images, calculates the confidence level of all points relative to the seabed boundary points for each cross-section of the port and starboard images. The point with the highest confidence level is extracted as the port and starboard seabed boundary for that cross-section. Obtaining the port and starboard boundary points for all cross-sections yields the seabed line of the strip image, which conforms to known prior knowledge and features. The seabed elevation information of the sonar platform relative to the seabed directly below is calculated based on the extracted seabed line. Using this elevation information as a basis, the slant range information in the strip image data can be further converted into horizontal range information, improving the accuracy of the side-scan sonar topographic images.

[0045] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0046] This embodiment provides a seabed elevation detection method based on prior knowledge from side-scan sonar images. See [link to previous document]. Figure 2 This includes the following steps:

[0047] Step 1: Obtain the port side stripe image and the starboard side stripe image, and convert them into grayscale images to obtain the port side grayscale matrix IL and the starboard side grayscale matrix IR.

[0048] The port and starboard side strip images are acquired using a two-dimensional side-scan sonar measurement platform and converted into grayscale images. The image data matrices are IL and IR, respectively. The size of the port grayscale matrix IL and the starboard grayscale matrix IR is M×N, where M is the number of points in a cross section and N is the number of cross sections.

[0049] Step 2: Based on a preset threshold, classify the port grayscale matrix IL and the starboard grayscale matrix IR respectively to obtain the port classification matrix CL and the starboard classification matrix CR.

[0050] For the port and starboard grayscale matrices, classification is performed based on a set threshold q. Pixels with grayscale values ​​greater than or equal to the threshold q are predicted targets and have a classification value of 1; pixels with grayscale values ​​less than the threshold q are predicted background and have a classification value of 0. This results in the port classification matrix CL and the starboard classification matrix CR.

[0051] Specifically, Where j∈[1, N] represents the section number (or row number), and i∈[1, M] represents the section point number (or column number, based on the sound source emission point, i.e., the rightmost point of the port side image, the leftmost point of the starboard side image, or as shown below). Figure 1 The centerline of the directly stitched image shown is the starting point; CL(i,j) represents the classification value of the element in the i-th row and j-th column of the port side classification matrix, IL(i,j) represents the gray value of the element in the i-th row and j-th column of the port side gray matrix, CR(i,j) represents the classification value of the element in the i-th row and j-th column of the starboard side classification matrix, and IR(i,j) represents the gray value of the element in the i-th row and j-th column of the starboard side gray matrix.

[0052] Step 3: Match each section in the port side classification matrix CL based on the port side template to obtain the port side correlation matrix XL; match each section in the starboard side classification matrix CR based on the starboard side template to obtain the starboard side correlation matrix XR.

[0053] For the port and starboard classification matrices CL and CR, using, for example Figure 2 The 1×7 size port and starboard templates shown are matched to each section to calculate their relevant values.

[0054] The port side template TL is: [1 1 1 1 0 0 0], the starboard side template TR is: [0 0 0 1 1 1 1], and the formula for calculating the port side correlation matrix XL is: The formula for calculating the starboard correlation matrix XR is as follows:

[0055] Where XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port side correlation matrix, TL(k) represents the value of the k-th element in the port side template, CL(i+k,j) represents the classification value of the element in the (i+k)-th row and j-th column of the port side classification matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard side correlation matrix, TR(k) represents the value of the k-th element in the starboard side template, and CR(i+k,j) represents the classification value of the element in the (i+k)-th row and j-th column of the starboard side classification matrix.

[0056] The following section provides further explanation of how to obtain the port and starboard correlation matrix.

[0057] From the sound source (i.e., the rightmost point of the port side image, the leftmost point of the starboard side image, or as...) Figure 1The sequence of cross-sectional points CL starting from the centerline of the directly stitched image shown. j (i) and CR j In (i), the regional features near the seabed boundary to be identified are: background (classification value 0) → target (i.e. seabed, classification value 1) → remain as target state (value 1), the gradient jumps from 0 to 1 and remains at 1.

[0058] In contrast, for continuous background regions, the gradient remains 0 as background (value 0) → background (value 0); for continuous target regions, the gradient remains 1 as target (value 1) → target (value 1).

[0059] Features of abrupt change points caused by noise and interference: background (value 0) → target (value 1) → background (value 0). After the gradient jumps from 0 to 1, there will be a negative gradient that causes the classification to return to the background, and then it will become 0 again, maintaining the background state.

[0060] This demonstrates that the classification changes at the seabed boundary exhibit unique characteristics. Using the aforementioned port and starboard templates, the regional features at the seabed boundary can be identified by matching the maximum correlation value, thus determining the seabed boundary location (the 4th point of the most correlated region, i.e., the point where the background transitions to the target). Furthermore, during template matching, the edge points on both sides of the cross-section lacked region values ​​on one side, resulting in the final correlation result sequence having 6 fewer points than the original cross-section. Therefore, for the classification value sequences of each cross-section in the port and starboard images, 3 points were extended on each side using edge values, ensuring the final correlation result sequence maintains a one-to-one correspondence with the original cross-section point count. Since the seabed location is not located on either side, the extension operation does not affect the results. This step yields the port side correlation matrix XL and the starboard side correlation matrix XR.

[0061] Step 4: Obtain the port side continuity matrix QL based on the port side correlation matrix XL, and obtain the starboard side continuity matrix QR based on the starboard side correlation matrix XR.

[0062] Taking the i-th point L(i,j) on section j of one side as the center and the corresponding i-th point L(i,j-1) on section j-1 of the same side as the center, find the maximum correlation value in the first neighborhood formed by the n points before and after it; obtain the continuity matrix of the side based on the maximum correlation value in the first neighborhood and the correlation matrix of the side.

[0063] Taking the left side as an example, the port side correlation sequence XL is calculated by the above steps. Taking the i-th point L(i,j) on the j-1 section of the port side and the corresponding i-th point L(i,j-1) on the next section j-1 of the port side as the center, the maximum correlation value in the neighborhood formed by the n points before and after is found, as expressed as follows:

[0064] Max(XL(in,j-1),XL(i-n+1,j-1),…,XL(i+n,j-1)).

[0065] The product of the maximum correlation value and XL(i,j) is used as the continuity evaluation value. The port continuity matrix QL is expressed as follows:

[0066] QL(i,j)=XL(i,j)·Max(XL(in,j-1),XL(i-n+1,j-1),…,XL(i+n,j-1)).

[0067] Where QL(i,j) represents the continuity evaluation value of the element in the i-th row and j-th column of the port side continuity matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port side correlation matrix, and n represents the number of points constituting the first neighborhood. n is determined based on the sailing speed and longitudinal resolution of the side-scan sonar platform, for example, n = 5.

[0068] Similarly, the starboard continuity matrix QR can be obtained, expressed as follows:

[0069] QR(i,j)=XR(i,j)·Max(XR(in,j-1),XR(i-n+1,j-1),...,XR(i+n,j-1)).

[0070] Where QR(i,j) represents the continuity assessment value of the element in the i-th row and j-th column of the starboard continuity matrix, and XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard correlation matrix.

[0071] Since the side-scan sonar platform measures the echo of a section at a certain location and then measures the echo of the next section as it navigates to the next location, with little interval between the two locations, it is reasonable to assume that the elevation of the sonar platform relative to the seabed does not change significantly. Therefore, the seabed boundary positions detected by adjacent sections will not be too far apart. In the above continuity calculation formula, based on the correlation value of its own point, multiplying it by the correlation value of the predicted seabed boundary point existing in the neighborhood of the corresponding point in the previous section, the continuity of the detected seabed boundary point echo along the section can be evaluated, which is helpful in filtering out erroneously detected seabed positions.

[0072] Step 5: Obtain the port side symmetry matrix SL based on the port side correlation matrix XL, and obtain the starboard side symmetry matrix SR based on the starboard side correlation matrix XR.

[0073] Taking the symmetrical point R(i,j) of the i-th point L(i,j) on the j-th section of one side of the ship and the i-th point R(i,j) on the j-th section of the other side of the ship as the center, find the maximum correlation value in the second neighborhood formed by the m points before and after it; based on the maximum correlation value in the second neighborhood and the correlation matrix of that side of the ship, obtain the symmetry matrix of that side. Taking the left side as an example, the port side correlation matrix XL is calculated by the above steps. Taking the symmetrical point R(ij) of the i-th point L(i,j) on the j-th section of the port side and the i-th point R(ij) on the j-th section of the starboard side as the center, find the maximum correlation value in the neighborhood formed by the m points before and after it, expressed as follows:

[0074] Max(XR(im,j),XR(i-m+1,j),…,XR(i+m,j)).

[0075] The product of the maximum correlation value and XL(i,j) is used as the symmetry evaluation value, and the port symmetry matrix SL is expressed as follows:

[0076] SL(i,j)=XL(i,j)·Max(XR(im,j), XR(i-m+1,j), ..., XR(i+m,j)).

[0077] Similarly, the starboard symmetry matrix SR can be obtained, expressed as follows:

[0078] SR(i,j)=XR(i,j)·Max(XL(im,j),XL(i-m+1,j),…,XL(i+m,j)).

[0079] Where SL(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the port symmetry matrix, SR(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the starboard symmetry matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the port correlation matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the starboard correlation matrix, and m represents the number of points constituting the second neighborhood. m is determined based on the measurement range and lateral resolution of the side-scan sonar, for example, m = 3.

[0080] The side-scan sonar platform emits sound waves to both sides, which are reflected back upon encountering the seabed. The echo intensity is recorded chronologically to obtain the side-scan sonar image. Based on this imaging mechanism, when sound waves are emitted to the port and starboard sides, the seabed directly below is usually the nearest seabed echo. The seabed boundary points on both sides correspond to the correct seabed location; therefore, the seabed echoes from both sides should arrive at the side-scan sonar simultaneously. In the side-scan sonar image, this is reflected as the port and starboard seabed boundaries being symmetrical relative to the centerline. In the above formula, the correlation value of the current point is multiplied by the correlation value of the predicted seabed boundary point existing in the neighborhood of the symmetrical location. This allows for the assessment of the port and starboard symmetry of the detected seabed boundary point echoes, which is helpful in filtering out incorrectly detected seabed locations.

[0081] Step 6: Combine the port side correlation matrix XL, the port side continuity matrix QL, and the port side symmetry matrix SL to obtain the port side confidence matrix BL; combine the starboard side correlation matrix XR, the starboard side continuity matrix QR, and the starboard side symmetry matrix SR to obtain the starboard side confidence matrix BR.

[0082] By combining the correlation matrix, symmetry matrix, and continuity matrix of the port and starboard images, a weighted composite confidence matrix is ​​synthesized. This matrix has the same size as the original side-scan sonar image. Each element in the confidence matrix corresponds to a pixel in the image, and the value of each element in the confidence matrix corresponding to a pixel in the image is the confidence level of the seabed boundary line in the port (or starboard) image.

[0083] The formula for calculating the port confidence matrix BL is: BL = (1 - k1 - k2)XL + k1SL + k2QL; the formula for calculating the starboard confidence matrix BR is: BR = (1 - k1 - k2)XR + k1SR + k2QR.

[0084] Where k1 represents the weighting factor for symmetry evaluation and k2 represents the weighting factor for continuity evaluation. k1 and k2 are estimated from actual navigation parameters. For example, k1 = 0.3 and k2 = 0.3 are set, while the correlation coefficient is 1 - k1 - k2 = 0.4.

[0085] Step 7: In the port side confidence matrix BL, for each section, mark the point with the largest confidence value as the seabed boundary point of that port side section, and use the curve connecting the seabed boundary points of all port side sections as the port side seabed line; in the starboard side confidence matrix BR, for each section, mark the point with the largest confidence value as the seabed boundary point of that starboard side section, and use the curve connecting the seabed boundary points of all starboard side sections as the starboard side seabed line.

[0086] In the confidence matrices BL and BR of the port and starboard images, for each section j, the point with the highest confidence value is marked as the seabed boundary point VL(j) and VR(j). The curve connecting the seabed boundary points of all sections on the port and starboard sides is the seabed line of the side-scan sonar image.

[0087] Step 8: Calculate the seabed elevation based on the starboard seabed line and the port seabed line.

[0088] The seabed elevation of the j-th section on the port side is expressed as: The seabed elevation of the j-th section on the starboard side is expressed as:

[0089] Where HL(j) represents the seabed elevation of the j-th section on the port side, HR(j) represents the seabed elevation of the j-th section on the starboard side, VL(j) represents the seabed boundary point of the j-th section on the port side, VR(j) represents the seabed boundary point of the j-th section on the starboard side, c represents the underwater sound speed, ΔT represents the sampling time of the side-scan sonar image data points, and f s This indicates the sampling frequency of the side-scan sonar image data points.

[0090] The elevation of each cross section can be calculated from the pixel distances VL(j) and VR(j) between the seabed boundary point and the starting point of the sound source at each cross section j.

[0091] In summary, this invention utilizes the original port and starboard strip images obtained from a side-scan sonar measurement platform to automatically extract the seabed line from the images, thereby measuring the elevation of the sonar platform relative to the seabed. This lays the foundation for the next step of converting slant range information into horizontal range information and synthesizing terrain images. This invention first uses a threshold to coarsely classify images, classifying each pixel as either a predicted target or predicted background (including noise, clutter, etc.). Then, based on the side-scan sonar imaging mechanism and its gradient features in the local seabed boundary region, seabed region, water body region, and clutter region, a template matching method is used to calculate the similarity between the neighborhood of each pixel and the features of the local seabed boundary region. Based on the correlation matrix formed by the template matching, the correlation values ​​of the seabed distribution on adjacent sections are compared to calculate its continuity matrix, revealing the continuity of the seabed distribution of each pixel with its corresponding points on adjacent sections. Based on the correlation matrix formed by the template matching, the correlation values ​​on the port and starboard sides are compared to calculate its symmetry matrix, revealing the symmetrical distribution of each pixel with its corresponding points at its symmetrical positions. Finally, by combining gradient correlation, continuity index, and symmetry index, and weighting and superimposing all prior information constraints, a comprehensive confidence index is synthesized. By detecting the maximum confidence index of each section, the most probable seabed boundary point is extracted to obtain the seabed line, thereby calculating the elevation of the side-scan sonar platform during the section measurement process using simple acoustic wave propagation. This invention makes full use of prior information such as gradient characteristics, symmetry, and continuity of the seabed boundary. It is simple to operate, robust, and has the potential for wide application.

[0092] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting seabed elevation from side scan sonar images based on prior knowledge, characterized by, The method comprises the following steps: Step 1, obtaining a left side strip image and a right side strip image, and converting the images into gray images to obtain a left side gray matrix IL and a right side gray matrix IR; Step 2, based on a preset threshold, classifying the left side gray matrix IL and the right side gray matrix IR respectively to obtain a left side classification matrix CL and a right side classification matrix CR; Step 3, matching each section in the left side classification matrix CL based on a left side template to obtain a left side correlation matrix XL, and matching each section in the right side classification matrix CR based on a right side template to obtain a right side correlation matrix XR; Step 4, obtaining a left side continuity matrix QL based on the left side correlation matrix XL, and obtaining a right side continuity matrix QR based on the right side correlation matrix XR; Step 5, obtaining a left side symmetry matrix SL based on the left side correlation matrix XL, and obtaining a right side symmetry matrix SR based on the right side correlation matrix XR; Step 6, synthesizing the left side correlation matrix XL, the left side continuity matrix QL and the left side symmetry matrix SL to obtain a left side confidence matrix BL by weighting, and synthesizing the right side correlation matrix XR, the right side continuity matrix QR and the right side symmetry matrix SR to obtain a right side confidence matrix BR by weighting; Step 7, in the left side confidence matrix BL, for each section, marking the point with the maximum confidence value as the sea bottom demarcation point of the section, and connecting the sea bottom demarcation points of all sections to obtain a left side sea bottom line; in the right side confidence matrix BR, for each section, marking the point with the maximum confidence value as the sea bottom demarcation point of the section, and connecting the sea bottom demarcation points of all sections to obtain a right side sea bottom line; Step 8, calculating a sea bottom elevation based on the right side sea bottom line and the left side sea bottom line.

2. The method of detecting seabed elevation from a side scan sonar image based on prior knowledge according to claim 1, wherein, In the step 1, the size of the left side gray matrix IL and the right side gray matrix IR is MxN; wherein M is the number of points of a section, and N is the number of sections; In the step 2, the pixel points with a gray value greater than or equal to the threshold value are divided into predicted targets, and the classification value is 1; the pixel points with a gray value less than the threshold value are divided into predicted backgrounds, and the classification value is 0; The left-hand side classification matrix CL is represented as: The right-hand side classification matrix CR is represented as: wherein q is the threshold value; i∈[1, M], i represents the section point serial number; j∈[1, N], j represents the section number; CL(i, j) represents the classification value of the element in the i-th row and the j-th column of the left side classification matrix, IL(i, j) represents the gray value of the element in the i-th row and the j-th column of the left side gray matrix, CR(i, j) represents the classification value of the element in the i-th row and the j-th column of the right side classification matrix, and IR(i, j) represents the gray value of the element in the i-th row and the j-th column of the right side gray matrix.

3. The method of detecting seabed elevation from a side scan sonar image based on prior knowledge according to claim 1, wherein, In the step 3, the left side template TL is [1 1 1 1 0 0 0], and the right side template TR is [0 0 0 1 1 1 1]. The calculation formula of the left side correlation matrix XL is: The calculation formula of the right side correlation matrix XR is: Wherein, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the left correlation matrix, TL(k) represents the value of the k-th element in the left template, CL(i+k,j) represents the classification value of the element in the i+k-th row and j-th column of the left classification matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the right correlation matrix, TR(k) represents the value of the k-th element in the right template, and CR(i+k,j) represents the classification value of the element in the i+k-th row and j-th column of the right classification matrix.

4. The method of claim 1, wherein, In step 4, the maximum correlation value in the first neighborhood formed by the previous and next n points is found with the i-th point L(i,j) on the j-th section of one side as the center on the corresponding i-th point L(i,j-1) on the j-1-th section of the side; The continuity matrix of the side is obtained based on the maximum correlation value in the first neighborhood and the correlation matrix of the side; The calculation formula of the left continuity matrix QL is: QL(i,j) = XL(i,j) Max(XL(i-n,j-1), XL(i-n+1,j-1), …, XL(i+n,j-1)); Wherein, QL(i,j) represents the continuity evaluation value of the element in the i-th row and j-th column of the left continuity matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the left correlation matrix, and n represents the number of points constituting the first neighborhood. The calculation formula of the right continuity matrix QR is: QR(i,j) = XR(i,j) Max(XR(i-n,j-1), XR(i-n+1,j-1), …, XR(i+n,j-1)); Wherein, QR(i,j) represents the continuity evaluation value of the element in the i-th row and j-th column of the right continuity matrix, and XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the right correlation matrix.

5. The prior-knowledge-based detection method of seabed elevation from side scan sonar images according to claim 1, wherein, In step 5, the maximum correlation value in the second neighborhood formed by the previous and next m points is found with the i-th point L(i,j) on the j-th section of one side as the center on the symmetric point, i.e. the i-th point R(i,j), on the j-th section of the other side; the symmetry matrix of the side is obtained based on the maximum correlation value in the second neighborhood and the correlation matrix of the side; The calculation formula of the left symmetry matrix SL is: SL(i,j) = XL(i,j) Max(XR(i-m,j), XR(i-m+1,j), …, XR(i+m,j)); The calculation formula of the right symmetry matrix SR is: SR(i,j) = XR(i,j) Max(XL(i-m,j), XL(i-m+1,j), …, XL(i+m,j)); Wherein, SL(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the left symmetry matrix, SR(i,j) represents the symmetry evaluation value of the element in the i-th row and j-th column of the right symmetry matrix, XL(i,j) represents the correlation value of the element in the i-th row and j-th column of the left correlation matrix, XR(i,j) represents the correlation value of the element in the i-th row and j-th column of the right correlation matrix, and m represents the number of points constituting the second neighborhood.

6. The prior-knowledge-based detection method of seabed elevation from side scan sonar images according to claim 1, wherein, In step 6, the calculation formula of the port confidence matrix BL is BL=(1-k1-k2)XL+k1SL+k2QL, and the calculation formula of the starboard confidence matrix BR is BR=(1-k1-k2)XR+k1SR+k2QR. Wherein, k1 represents a weight factor of symmetry evaluation, and k2 represents a weight factor of continuity evaluation.

7. The prior-knowledge-based detection method of seabed elevation from side scan sonar images according to claim 1, wherein, In step 8, the sea floor elevation of the jth cross section on the port side is represented as: The sea floor elevation of the jth cross section on the starboard side is represented as: wherein HL(j) represents the sea bottom elevation of the jth section of the port side, HR(j) represents the sea bottom elevation of the jth section of the starboard side, VL(j) represents the pixel distance of the position of the sea bottom demarcation point of the jth section of the port side from the start of the sound source position, VR(j) represents the pixel distance of the position of the sea bottom demarcation point of the jth section of the starboard side from the start of the sound source position, c represents the underwater sound speed, ΔT represents the sampling time of the side scan sonar image data point, f s represents the sampling frequency of the side scan sonar image data point.

Images