A method and system for constructing a bridge scour pattern map based on underwater sonar point cloud

Abstract

The application discloses a bridge scouring morphology atlas construction method and system based on underwater sonar point cloud, comprising the following steps: obtaining underwater sonar point cloud data in a preset sampling range, preprocessing to obtain preprocessed point cloud; performing semantic pre-segmentation on the preprocessed point cloud to obtain a riverbed point set, determining a reference bed surface model based on the riverbed point set to obtain a non-negative scouring depth field; extracting a scouring candidate point set, determining a main scouring pit area, and calculating a measurement feature set; generating a morphology semantic unit in the main scouring pit area; based on the main scouring pit area, the measurement feature set and the morphology semantic unit, a bridge scouring morphology atlas is constructed, the bridge scouring morphology atlas comprising a node set, an edge set, node attributes, edge attributes, traceability information, quality information and graph-level metadata. The application realizes the structured expression, traceable review and engineering interpretation of the bridge scouring morphology in the underwater sonar point cloud.

Description

Technical Field

[0001] This invention belongs to the field of bridge knowledge graph construction technology, and in particular relates to a method and system for constructing bridge scour morphology graphs based on underwater sonar point clouds. Background Technology

[0002] The riverbed surrounding bridge foundations is prone to localized scouring under the influence of water flow, forming scour pits with typical morphological structures such as "pit bottom - pit slope - pit edge - main scour axis". In engineering practice, underwater sonar is often used to obtain riverbed point cloud or gridded water depth data for scour depth assessment and risk warning.

[0003] In existing technologies, the processing of sonar point clouds mainly focuses on 3D surface reconstruction, mesh filtering, contour line extraction, or local geometric feature calculation. The output results are usually the original point cloud, depth field, extreme points, or finite geometric measurements. Although these results can reflect the degree of local erosion, they still have the following problems: 1) Lack of structured representation for hydraulic mechanisms: It is difficult to organize morphological semantic units such as pit bottom, pit slope, pit edge, valley line skeleton and deepest point with a unified data structure, and it is even more difficult to express the subordinate relationship, adjacency relationship, order relationship and topological connectivity relationship between them; 2) Lack of traceability: Engineering scenarios often require answering "which original points support a certain morphological node, what criteria were used to generate it, and what is its quality", but existing methods mostly lack chain evidence that traces back from the graph node to the point cloud index; 3) Lack of comparability across time periods and work sites: The scour morphology of bridges varies significantly under different working conditions, and relying solely on grids or local indicators is not conducive to long-term comparative analysis, retrieval, and knowledge accumulation. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a method and system for constructing bridge scour morphology maps based on underwater sonar point clouds, so as to realize the structured expression, traceable verification and engineering interpretation of bridge scour morphology in underwater sonar point clouds.

[0005] Technical Solution: To achieve the above objectives, the present invention provides a method for constructing bridge scour morphology maps based on underwater sonar point clouds, comprising the following steps: (1) Using the center of the bridge pier as a reference, acquire underwater sonar point cloud data within a preset sampling range. For the underwater sonar point cloud data Preprocessing is performed to obtain a preprocessed point cloud. ; (2) The preprocessed point cloud Semantic pre-segmentation is performed to obtain the riverbed point set. and establish riverbed point set underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number in the middle; (3) Based on riverbed point set Determine a reference bed model, and calculate the riverbed point set based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area Calculate the measurement feature set, which includes the deepest anchor point, maximum scour depth, sequence of representative points on the pit edge, pit edge depth level, area and volume measurement information, and maximum slope measurement information. (4) In the main scour pit area Inside, based on the non-negative scour depth field, the sequence of representative points on the pit edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the pit edge ring, the pit bottom point set, the pit slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded. (5) Based on the main scour pit area The bridge scour morphology map is constructed by measuring feature sets and morphological semantic units. The bridge scour morphology map includes node sets, edge sets, node attributes, edge attributes, source information, quality information and graph-level metadata.

[0006] Optionally, the preset sampling range in step (1) is determined by a preset sampling radius. Confirmed; the underwater sonar point cloud data This includes the three-dimensional coordinate information of each point. When the sonar device outputs echo intensity information and sampling time information, the echo intensity information and sampling time information are recorded as point attributes; this is the underwater sonar point cloud data. The preprocessing steps of outlier removal, statistical denoising, voxel downsampling, coordinate unification, and normalization are performed sequentially to obtain the preprocessed point cloud. It generates preprocessed metadata including coordinate system, units, normalization method, normalization parameters, statistical denoising parameters, and voxel downsampling parameters.

[0007] Optionally, in step (2), the preprocessed point cloud... Semantic pre-segmentation is performed, and the preprocessed point cloud is segmented based on point cloud normals, local flatness, elevation distribution, and spatial connectivity features. Divided into riverbed point sets and bridge foundation point set Among them, riverbed point set The set of points belonging to the riverbed surface, and the set of bridge foundation points. It is the set of points that belong to the surface of the pier or bridge foundation entity.

[0008] Optionally, step (3) specifically includes the following steps: (3.1) For the riverbed point set Robust fitting is performed to obtain a reference bed model, which serves as the riverbed elevation benchmark for calculating scour depth. The reference bed model adopts a local planar model, and its expression is: , in, , , For reference bed surface parameters, , Here are the plane coordinates of the riverbed point. Predict the elevation of the reference bed surface corresponding to the planar position; at the same time, record the proportion of fitted in-points, the statistics of the fitted residuals and the parameters of the reference bed surface to form the reference bed surface fitting information. (3.2) For the riverbed point set any riverbed point The scour depth of any riverbed point relative to the reference bed surface is calculated based on the reference bed surface model. The calculation formula is: , in, Riverbed point The measured elevation; when the calculated scour depth value is less than zero, the scour depth value of the riverbed point is counted as zero; from the riverbed point set The non-negative scour depth field is composed of the non-negative scour depth values ​​corresponding to each riverbed point. Then, based on the non-negative scour depth field, the riverbed point set is... Scour candidate points are determined for each riverbed point, and a scour candidate point set is extracted. Specifically, using riverbed points For the target riverbed point, its neighborhood is determined. The riverbed point neighborhood is the set of neighboring points centered on the target riverbed point, determined by a search distance parameter. Based on the elevation difference or scour depth difference between the target riverbed point and its neighboring riverbed points, the elevation undulation characteristics of the target riverbed point are obtained. Then, a local ternary model algorithm is used to encode the elevation undulation characteristics to obtain the scour confidence value of the target riverbed point. Riverbed points with scour confidence values ​​not less than a preset confidence threshold are classified as scour candidate points, and riverbed points with scour confidence values ​​less than the preset confidence threshold are classified as non-scour points. All scour candidate points form a scour candidate point set. ; (3.3) Set of candidate scour points Determining the main scour pit area using a depth threshold method Specifically, it refers to: using the scouring candidate point set The scour depth values ​​constitute a scour depth sample set, and the high quantile of the scour depth sample set is calculated. , median and median absolute deviation and determine the depth threshold. : , in, To preset the high quantile parameters, To achieve the lower threshold of the scour depth engineering, This is the relative threshold coefficient. These are robust statistical coefficients; scour depth value Not less than the depth threshold Points with the deepest points are selected as depth candidates, and spatial clustering is performed based on the planar coordinates of these candidates to obtain several candidate clusters. Among these clusters, the cluster containing the deepest point is selected as the main scour pit area. ; (3.4) Main scour pit area The point on the riverbed with the largest scour depth is selected as the deepest point anchor point, thus obtaining the location of the deepest point and the maximum scour depth. satisfy: , in, Indicates riverbed point The scour depth value, (3.5) Using the deepest anchor point as the origin of the polar coordinates, the main scour pit area is... Calculate the polar coordinates of any point on the riverbed: , , in, The coordinates of the plane at the origin of the polar coordinate system are... The distance from the riverbed point to the origin of the polar coordinate system is the planar distance. For the corresponding azimuth angle; According to the preset azimuth angle, the box is divided into steps. The azimuth angle is divided into several azimuth bins. Within each azimuth bin, riverbed points with scour depth values ​​not exceeding the candidate depth threshold for the pit edge are selected as candidate pit edge points. The candidate depth threshold for the pit edge is determined by the maximum scour depth and a preset proportional coefficient. Subsequently, the pit edge radius of that azimuth bin is determined by the high quantile of the radius of the candidate pit edge points. : , in, For the first The set of candidate points for the pit edge within each azimuth angle bin. For quantile operators, The high resolution parameters are preset; then, within this azimuth caliber, the radius of the pit edge is selected. The closest point is used as the representative point of the pit edge, forming a sequence of representative points of the pit edge ordered by azimuth. Pit edge depth horizontal The average value of the scour depth at the representative point on the edge of the pit is taken, based on the statistical values ​​of the scour depth at the representative point on the edge of the pit. (3.6) The non-negative scour depth field is represented by a grid according to the preset grid resolution to obtain the gridded scour depth field, with the scour depth not less than the pit edge depth. The grid cells are used as the projection area of ​​the scour pit, and the projected surface area and volume of the scour pit are calculated. Let the grid cell depth be The area of ​​the grid cell is The projected surface area of ​​the scour pit satisfy: , in, This is an indicator function; it takes the value 1 if the condition inside the parentheses is true, and 0 otherwise. Based on the depth level of the pit edge Calculate the volume of the scour pit : , After the calculation is completed, area and volume measurement information is generated; the area and volume measurement information includes projected surface area, scour pit volume, grid resolution, pit edge depth level, and grid cell indexes involved in the calculation; (3.7) Calculate the maximum slope, the location of the maximum slope, and the direction of the slope within the projected area of ​​the scour pit. The local bed elevation is expressed as: , in, To obtain the slope value, the gradient of the local bed elevation is calculated for the scour depth value at the corresponding planar location: , Within the projected area of ​​the scour pit, select the point or grid cell with the largest slope value as the location of the maximum slope, and take the slope value corresponding to this location as the maximum slope; the slope direction is taken as the direction of the steepest descent of the bed surface, and its direction vector is: , Simultaneously, the azimuth angle of the slope direction is recorded. After the calculation is completed, the maximum slope measurement information is generated. The maximum slope measurement information includes the maximum slope value, the location of the maximum slope, the slope direction, the slope azimuth angle, and the neighboring point index or grid cell index that participated in the slope calculation.

[0009] Optionally, step (4) specifically includes the following steps: (4.1) Based on the sequence of representative pit edge points obtained in step (3), connect adjacent representative pit edge points in ascending order of azimuth angle to form a candidate pit edge loop; for adjacent representative pit edge points and Calculate the azimuth notch and spatial distance : , , in, The sequence index is used to represent the pit edge points; when determining the first and last adjacent points, the last pit edge point in the sequence is connected to the first pit edge point in the sequence, and the azimuth is then used. Periodic compensation; When the adjacent pit edge representative points satisfy ,and At that time, establish the connectivity between representative points on adjacent pit edges; among which, The angle gap threshold is determined based on the azimuth box step size. The spatial distance gap threshold is determined based on the average spacing of the point cloud or the grid resolution. Calculate the maximum angular gap and maximum spatial gap of the candidate ring at the pit edge: , , when and When the crater edge is closed, the sequence of representative points is closed into a crater edge loop; otherwise, it is retained as an unclosed crater edge, and the crater edge closure status, the maximum gap and its corresponding azimuth angle interval are recorded as crater edge quality information; the maximum gap includes the maximum angular gap and the maximum spatial gap; (4.2) Using the deepest anchor point as the radial analysis center, construct the scour depth as a function of radius within each azimuth angle box. A changing radial profile refers to a one-dimensional function in which the scour depth of a riverbed point varies with its planar distance to the deepest anchor point within the same azimuth box. For each radial profile, calculate the first and second order variations of the scour depth, and then determine the boundary radius between the pit bottom and the pit slope based on the location of the abrupt change in depth. Specifically, this involves constructing a fracture scoring function: , in, For the first Radial profile within an azimuth angle sub-box For first-order changes, It is a second-order change. For curvature surrogate quantity, symbol " "Indicates normalization processing, , , These are the preset weighting coefficients for the fracture scoring function; The location where the fracture score function reaches its maximum value is taken as the pit bottom-pit slope boundary radius in that azimuth direction. : , Record the pit bottom-pit slope boundary radius and the fracture scoring function weight coefficient in each azimuth direction as the pit bottom-pit slope boundary parameter; (4.3) For the main scour pit area For any point within the deepest point, calculate its radius relative to the deepest anchor point. and azimuth ;when When, the point is assigned to the pit bottom point set; when At that time, the point is assigned to the pit slope point set, where The radius of the pit edge at the corresponding azimuth angle ; Perform connectivity filtering on the pit bottom point set, and retain the connected component with the most points or the largest area as the final pit bottom point set; the remaining scattered points are not considered as the main body of the pit bottom, and the pit slope point set consists of the points between the pit bottom boundary and the pit edge; record the pit bottom connectivity filtering results and the support point indexes corresponding to the pit bottom point set and the pit slope point set. (4.4) In the main scour pit area When generating the valley line skeleton, first determine the main direction of the valley line. Specifically, perform principal component analysis on the representative points of the pit edge and take the direction corresponding to the largest eigenvalue as the main direction of the valley line. When the number of representative points at the edge of the pit is less than the preset threshold, the least squares straight line fitting direction of the valley line candidate points is used to replace the main direction; the source of the main direction of the valley line is recorded. (4.5) Preset the strip width and divide the main scour pit area along the main direction of the valley line. Divided into several strips, for the main scour pit area For any point within the valley, its projected coordinates along the principal direction of the valley line. for: , in, Here are the plane coordinates of this point. The plane coordinates of the deepest anchor point. The main direction of the valley line; Within each zone, the point with the greatest scour depth is selected as the representative point of the valley line. : , in, For the first If the number of points in a certain strip is less than the preset threshold for the number of points in a strip, the strip width is increased or the strip is skipped, and the number of points represented by the valley line, the number of valid strips, and the number of missing strips are recorded. (4.6) Connect the representative points of each valley line in ascending order of projected coordinates to form a valley line skeleton; the valley line connection relationship between adjacent valley line representative points is used to represent the order relationship of the valley line skeleton, and record the spatial distance, strip number and support point index between adjacent valley line representative points; after the valley line skeleton is generated, output the valley line representative point sequence, valley line connection relationship, valley line main direction and valley line quality information; among which, the valley line quality information includes the source of the valley line main direction, the number of effective strips, the number of missing strips and the number of valley line representative points; (4.7) Output morphological semantic units, which include pit edge rings, pit bottom point sets, pit slope point sets, and valley line skeletons. The deepest point anchor point serves as the unified positioning reference for morphological semantic units. At the same time, output the support point index, generation parameters, and quality information corresponding to the morphological semantic units. The generation parameters include azimuth box step size, angle gap threshold, spatial distance gap threshold, pit bottom-pit slope boundary parameters, strip width, and strip point number threshold. The quality information includes pit edge quality information, pit bottom connectivity screening results, and valley line quality information.

[0010] Optionally, step (5) specifically includes the following steps: (5.1) Based on the main scour pit area A node set is constructed from morphological semantic units and measurement feature sets. The node set includes scour pit object nodes, morphological semantic unit nodes, topological point nodes, and measurement feature nodes. Scour pit object nodes refer to the nodes formed by the main scour pit area. The generated corresponding nodes are used to represent a main scour pit as a whole; Morphological semantic unit nodes refer to four types of corresponding nodes generated from the pit bottom point set, pit slope point set, pit edge loop, and valley line skeleton, respectively. Topological point nodes refer to the two corresponding nodes generated by the pit edge representative point and the valley line representative point, respectively. The measurement feature nodes refer to the four corresponding nodes generated by the deepest point anchor point, projected surface area, scour pit volume, and maximum slope. Each node records the node type, geometric location, and corresponding support point index or support grid cell index; the support point index is used to record the original point number of the riverbed point corresponding to the node in the underwater sonar point cloud data; (5.2) For each morphological semantic unit node, its geometric position is determined according to the point set corresponding to the node, and the centroid of the support point set is used as the geometric position of the node. : , in, Represents a node. Represents a node The corresponding support point set, Indicates the riverbed points in the support point set; For nodes generated from a point set, the node depth attribute is also determined based on the scour depth values ​​of each riverbed point in the support point set. Specifically, the median of the scour depth values ​​of each riverbed point in the support point set is used as the node depth attribute. : , in, Indicates riverbed point The corresponding scour depth value is used to write the node's geometric position and node depth attributes into the node attributes. For nodes generated from measurement features, the corresponding measurement features are directly written into the node attributes; among them, the corresponding node generated from the deepest point anchor point records the location of the deepest point and the maximum scour depth, the corresponding node generated from the projected surface area records the projected surface area of ​​the scour pit, the corresponding node generated from the scour pit volume records the volume of the scour pit, and the corresponding node generated from the maximum slope records the maximum slope value, the location of the maximum slope, the slope direction, and the slope azimuth. (5.3) Construct an edge set based on the hierarchical, subordinate, sequential, and adjacency relationships between nodes; Hierarchical relationship edges are used to represent the inclusion relationship between scour pit object nodes and morphological semantic unit nodes; for each morphological semantic unit node, a hierarchical relationship edge is established between it and the scour pit object node. Subordination edges are used to indicate that measurement feature nodes belong to the same main scour pit object; for the deepest point anchor node, projected surface area node, volume node and maximum slope node, respectively establish subordination edges between them and the scour pit object node; Sequential relationship edges are used to represent the arrangement order between topological point nodes; for pit edge topological point nodes, sequential relationship edges are established in order of azimuth angle from small to large; for valley line topological point nodes, sequential relationship edges are established in order of projected coordinates in the main direction of the valley line. Adjacency edges are used to represent the spatial adjacency relationships between morphological semantic units; an adjacency edge is established when the nearest distance between the support point sets corresponding to two morphological semantic units is not greater than a preset adjacency distance threshold. , in, and These are two morphological semantic unit nodes. For morphological semantic unit nodes The corresponding support point set, To support point sets points within, For morphological semantic unit nodes The corresponding support point set, To support point sets points within; To preset the adjacency distance threshold, it is determined based on the average point spacing or grid resolution of the riverbed point set, typically taken as a fraction of the average point spacing or the diagonal length of the grid cell. times; (5.4) Connect the topological point nodes generated from the representative points of the pit edge in azimuth order to form the pit edge topological structure; when the sequence of representative points of the pit edge satisfies and Establish closed edges at both ends, and record the pit edge closure status, maximum angle gap, maximum spatial gap, angle gap threshold, and spatial distance gap threshold in the edge attributes; The topological point nodes generated for the valley line representative points are connected in the order of their projected coordinates along the main direction of the valley line to form the valley line skeleton topological structure; the strip number, spatial distance, and support point index of adjacent valley line representative points are recorded in the edge attributes. (5.5) Write source information for nodes and edges. Node source information includes node source, supporting point index, supporting grid cell index and generation parameters; edge source information includes edge type, edge generation basis, related supporting point pairs and related topological point sequence. Among them, the source information of the morphological semantic unit node comes from the support point index; the source information of the corresponding node generated by the projected surface area and the corresponding node generated by the scour pit volume comes from the grid cell index participating in the area and volume calculation; the source information of the corresponding node generated by the maximum slope comes from the maximum slope location and its neighboring point index or neighboring grid cell index. Establish a correspondence between bridge scour morphology maps, riverbed point sets, and underwater sonar point cloud data by tracing the source information; (5.6) Record the number of nodes and the number of edges as quality information, and write quality information into the bridge scour morphology map. The quality information includes the quality information of the pit edge, the screening results of the pit bottom connectivity, the quality information of the valley line, the number of nodes and the number of edges. When the pit edge is not closed, the pit bottom connectivity is insufficient, the number of effective valley strips is less than the preset threshold, or the nodes and edges do not meet the consistency verification requirements, the corresponding abnormal marker is recorded in the quality information. (5.7) Combine the node set, edge set, node attributes, edge attributes, traceability information and quality information to form a bridge scour morphology map. The bridge scour morphology map meets the preset map data mode. The map-level metadata includes map identifier, coordinate system, unit, center of main scour pit, node type, edge type, number of nodes, number of edges, upstream data source and processing parameters. Before output, a consistency check is performed on the bridge scour morphology map. The consistency check includes checking the uniqueness of node identifiers, the existence of edge endpoints, the legality of node types, and the legality of edge types. After passing the consistency check, the bridge scour morphology map is output.

[0011] The bridge scour morphology map construction system based on underwater sonar point clouds of the present invention includes: The preprocessed point cloud acquisition module is used to acquire underwater sonar point cloud data within a preset sampling range, with the center of the bridge pier as the reference. For the underwater sonar point cloud data Preprocessing is performed to obtain a preprocessed point cloud. ; The riverbed point set acquisition module is used to process the preprocessed point cloud. Semantic pre-segmentation is performed to obtain the riverbed point set. and establish riverbed point set underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number in the middle; The module for calculating measurement features is used to calculate the features based on the riverbed point set. Determine a reference bed model, and calculate the riverbed point set based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area Calculate the measurement feature set, which includes the deepest anchor point, maximum scour depth, sequence of representative points on the pit edge, pit edge depth level, area and volume measurement information, and maximum slope measurement information. The morphological semantic unit module is used to generate morphological semantic units in the main scour pit area. Inside, based on the non-negative scour depth field, the sequence of representative points on the pit edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the pit edge ring, the pit bottom point set, the pit slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded. The map construction module is used to construct maps based on the main scour pit area. The bridge scour morphology map is constructed by measuring feature sets and morphological semantic units. The bridge scour morphology map includes node sets, edge sets, node attributes, edge attributes, source information, quality information and graph-level metadata.

[0012] The electronic device of the present invention includes a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the bridge scour morphology map construction method based on underwater sonar point clouds as described above.

[0013] The present invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for constructing a bridge scour morphology map based on underwater sonar point clouds as described above.

[0014] The present invention provides a computer program product, comprising a computer program that, when executed by a processor, implements the steps of a bridge scour morphology map construction method based on underwater sonar point clouds as described above.

[0015] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: (1) Based on underwater sonar point clouds, this invention can robustly extract the main scour pit area and its key morphological semantic units under common conditions such as noise, missing measurements, obstruction and sparse point clouds. (2) This invention realizes the structured expression of bridge scour morphology by organizing fixed enumeration morphological semantic units such as pit edge, pit bottom, pit slope, valley line skeleton and deepest point anchor point; (3) This invention can explicitly express the hierarchical, adjacency, sequence and topological relationships within a scour pit through the unified organization of nodes, edges, attributes, metadata, traceability fields and quality fields; (4) By writing support point indexes, processing parameter sources and intermediate result references to graph nodes and edges, this invention forms a traceable evidence chain of “point cloud - intermediate result - graph”, which is convenient for subsequent verification, retrieval and engineering interpretation. (5) The morphological maps output by this invention have a unified data format standard, which facilitates comparative analysis and engineering applications across time periods, work sites and datasets. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the implementation of the present invention; Figure 2 This is a schematic diagram of the scour depth field construction and main scour pit region extraction of the present invention; Figure 3 This is a schematic diagram illustrating the pit edge extraction, ring connectivity construction, and maximum gap determination of the present invention; Figure 4 This is a schematic diagram illustrating the extraction of the pit bottom and the determination of the pit bottom-pit slope boundary line in this invention; Figure 5 This is a schematic diagram of the data structure and node-edge relationship of the bridge scour morphology map of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0018] Example 1: As Figure 1 As shown, a method for constructing bridge scour morphology maps based on underwater sonar point clouds includes the following steps: (1) Using the center of the bridge pier as a reference, acquire underwater sonar point cloud data within a preset sampling range. The preset sampling range is determined by the preset sampling radius. Confirmed; the underwater sonar point cloud data This includes the three-dimensional coordinate information of each point. When the sonar device outputs echo intensity information and sampling time information, the echo intensity information and sampling time information are recorded as point attributes; this is the underwater sonar point cloud data. The preprocessed point cloud is obtained by sequentially performing outlier removal, statistical denoising, voxel downsampling, coordinate unification, and normalization. It generates preprocessed metadata including coordinate system, units, normalization method, normalization parameters, statistical denoising parameters, and voxel downsampling parameters.

[0019] (2) The preprocessed point cloud Semantic pre-segmentation is performed, and the preprocessed point cloud is segmented based on point cloud normals, local flatness, elevation distribution, and spatial connectivity features. Divided into riverbed point sets and bridge foundation point set Among them, riverbed point set The set of points belonging to the riverbed surface, and the set of bridge foundation points. This is the set of points belonging to the surface of the bridge piers or bridge foundations; and the riverbed point set is also established. underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number.

[0020] (3) Based on riverbed point set A reference bed model is determined to represent the riverbed elevation datum under scour conditions; the riverbed point set is calculated based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area The measurement feature set is calculated, including the deepest anchor point, maximum scour depth, sequence of representative points at the scour rim, scour rim depth level, area and volume measurement information, and maximum slope measurement information; the main scour crater area. The measurement feature set is used for subsequent morphological semantic unit generation and morphological map construction.

[0021] like Figure 2 As shown, step (3) specifically includes the following steps: (3.1) For the riverbed point set Robust fitting is performed to obtain a reference bed model, which serves as the riverbed elevation benchmark for calculating scour depth. The reference bed model adopts a local planar model, and its expression is: , in, , , For reference bed surface parameters, , Here are the plane coordinates of the riverbed point. The elevation of the reference bed surface corresponding to the planar position is predicted; at the same time, the proportion of fitted inliers, the statistics of the fitted residuals and the parameters of the reference bed surface are recorded to form the reference bed surface fitting information.

[0022] (3.2) For the riverbed point set any riverbed point The scour depth of any riverbed point relative to the reference bed surface is calculated based on the reference bed surface model. The calculation formula is: , in, Riverbed point The measured elevation; when the calculated scour depth value is less than zero, the scour depth value of the riverbed point is counted as zero; from the riverbed point set The non-negative scour depth field is composed of the non-negative scour depth values ​​corresponding to each riverbed point. Then, based on the non-negative scour depth field, the riverbed point set is... Scour candidate points are determined for each riverbed point, and a scour candidate point set is extracted. Specifically, using riverbed points For the target riverbed point, its neighborhood is determined. The riverbed point neighborhood is the set of neighboring points centered on the target riverbed point, determined by a search distance parameter. Based on the elevation difference or scour depth difference between the target riverbed point and its neighboring riverbed points, the elevation undulation characteristics of the target riverbed point are obtained. Then, a local ternary model algorithm is used to encode the elevation undulation characteristics to obtain the scour confidence value of the target riverbed point. Riverbed points with a scour confidence value not less than a preset confidence threshold of 0.05 are classified as scour candidate points, and riverbed points with a scour confidence value less than the preset confidence threshold are classified as non-scour points. All scour candidate points form a scour candidate point set. The search distance parameter, skip radius parameter, and ternary threshold parameter in the local ternary mode algorithm are determined by the dung beetle optimization algorithm.

[0023] (3.3) Set of candidate scour points Determining the main scour pit area using a depth threshold method Specifically, it refers to: using the scouring candidate point set The scour depth values ​​constitute a scour depth sample set, and the high quantile of the scour depth sample set is calculated. , median and median absolute deviation and determine the depth threshold. : , in, The preset high quantile parameter has a value range of 0.90 to 0.99; The lower limit threshold for scour depth is used to exclude shallow false scour points caused by measurement noise or local small undulations. This is the relative threshold coefficient. These are robust statistical coefficients; scour depth value Not less than the depth threshold Points with the deepest points are selected as depth candidates, and spatial clustering is performed based on the planar coordinates of these candidates to obtain several candidate clusters. Among these clusters, the cluster containing the deepest point is selected as the main scour pit area. .

[0024] (3.4) Main scour pit area The point on the riverbed with the largest scour depth is selected as the deepest point anchor point, thus obtaining the location of the deepest point and the maximum scour depth. satisfy: , in, Indicates riverbed point The scour depth value is used as the anchor point of the deepest point, which serves as the unified geometric benchmark for subsequent pit edge extraction, pit bottom and slope decomposition, and map node positioning.

[0025] (3.5) Using the deepest anchor point as the origin of the polar coordinates, the main scour pit area is... Calculate the polar coordinates of any point on the riverbed: , , in, The coordinates of the plane at the origin of the polar coordinate system are... The distance from the riverbed point to the origin of the polar coordinate system is the planar distance. For the corresponding azimuth angle; According to the preset azimuth angle, the box is divided into steps. The azimuth angle is divided into several azimuth bins. Within each azimuth bin, riverbed points with scour depth values ​​not exceeding the candidate depth threshold for the scour edge are selected as candidate scour edge points. The candidate depth threshold for the scour edge is determined by the maximum scour depth and a preset proportional coefficient, preferably 0% to 3% of the maximum scour depth. Subsequently, the scour edge radius of that azimuth bin is determined by the high quantile of the radius of the candidate scour edge points. : , in, For the first The set of candidate points for the pit edge within each azimuth angle bin. For quantile operators, To preset the high-resolution parameter, the value range is 0.90 to 0.99; then, within this azimuth angle bisector, a value is selected that is equal to the radius of the pit edge. The closest point is used as the representative point of the pit edge, forming a sequence of representative points of the pit edge ordered by azimuth. Pit edge depth horizontal The scour depth is determined by statistical values ​​of the representative points at the edge of the pit, and the average value of the scour depth at the representative points at the edge of the pit is taken.

[0026] (3.6) The non-negative scour depth field is represented by a grid according to the preset grid resolution to obtain the gridded scour depth field, with the scour depth not less than the pit edge depth. The grid cells are used as the projection area of ​​the scour pit, and the projected surface area and volume of the scour pit are calculated. Let the grid cell depth be The area of ​​the grid cell is The projected surface area of ​​the scour pit satisfy: , in, This is an indicator function; it takes the value 1 if the condition inside the parentheses is true, and 0 otherwise. Based on the depth level of the pit edge Calculate the volume of the scour pit : , After the calculation is completed, area and volume measurement information is generated. The area and volume measurement information includes the projected surface area, the volume of the scour pit, the grid resolution, the depth level of the pit edge, and the index of the grid cells involved in the calculation, which are used as the data source for the projected surface area nodes and volume nodes in the subsequent map.

[0027] (3.7) Calculate the maximum slope, the location of the maximum slope, and the direction of the slope within the projected area of ​​the scour pit. The local bed elevation is expressed as: , in, To obtain the slope value, the gradient of the local bed elevation is calculated for the scour depth value at the corresponding planar location: , Within the projected area of ​​the scour pit, select the point or grid cell with the largest slope value as the location of the maximum slope, and take the slope value corresponding to this location as the maximum slope; the slope direction is taken as the direction of the steepest descent of the bed surface, and its direction vector is: , Simultaneously, the azimuth angle of the slope direction is recorded. After the calculation is completed, the maximum slope measurement information is generated. The maximum slope measurement information includes the maximum slope value, the location of the maximum slope, the slope direction, the slope azimuth angle, and the neighbor point index or grid cell index that participated in the slope calculation. This information is used as the data source for the maximum slope node in the subsequent map.

[0028] (4) In the main scour pit area Within the system, based on the non-negative scour depth field, the sequence of representative points on the crater edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the crater edge ring, the crater bottom point set, the crater slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded.

[0029] like Figure 3 As shown, step (4) specifically includes the following steps: (4.1) Based on the sequence of representative pit edge points obtained in step (3), connect adjacent representative pit edge points in ascending order of azimuth angle to form a candidate pit edge loop; for adjacent representative pit edge points and Calculate the azimuth notch and spatial distance : , , in, The sequence index is used to represent the pit edge points; when determining the first and last adjacent points, the last pit edge point in the sequence is connected to the first pit edge point in the sequence, and the azimuth is then used. Periodic compensation; When the adjacent pit edge representative points satisfy ,and At that time, establish the connectivity between representative points on adjacent pit edges; among which, The angle gap threshold is determined based on the azimuth box step size. The spatial distance gap threshold is determined based on the average spacing of the point cloud or the grid resolution. Calculate the maximum angular gap and maximum spatial gap of the candidate ring at the pit edge: , , when and When the crater edge representative point sequence is closed into a crater edge loop, it is otherwise retained as an unclosed crater edge. The crater edge closure status, the maximum gap and its corresponding azimuth angle interval are recorded as crater edge quality information. The maximum gap includes the maximum angular gap and the maximum spatial gap.

[0030] (4.2) Using the deepest anchor point as the radial analysis center, construct the scour depth as a function of radius within each azimuth angle box. A changing radial profile refers to a one-dimensional function in which the scour depth of a riverbed point varies with its planar distance to the deepest anchor point within the same azimuth box. like Figure 4 As shown, the first-order and second-order variations of the scour depth are calculated for each radial profile, and the radius of the pit bottom-pit slope boundary is determined based on the location of the abrupt change in depth. Specifically, this involves constructing a fracture scoring function: , in, For the first Radial profile within an azimuth angle sub-box For first-order changes, It is a second-order change. For curvature surrogate quantity, symbol " "Indicates normalization processing, , , These are the preset weighting coefficients for the fracture scoring function; The location where the fracture score function reaches its maximum value is taken as the pit bottom-pit slope boundary radius in that azimuth direction. : , The pit bottom-pit slope boundary radius and the fracture scoring function weight coefficient are recorded in each azimuth direction as the pit bottom-pit slope boundary parameters.

[0031] (4.3) For the main scour pit area For any point within the deepest point, calculate its radius relative to the deepest anchor point. and azimuth ;when When, the point is assigned to the pit bottom point set; when At that time, the point is assigned to the pit slope point set, where The radius of the pit edge at the corresponding azimuth angle ; Connectivity filtering is performed on the pit bottom point set, and the connected component with the most points or the largest area is retained as the final pit bottom point set; the remaining scattered points are not considered as the main body of the pit bottom, and the pit slope point set consists of the points between the pit bottom boundary and the pit edge; the pit bottom connectivity filtering results and the support point indexes corresponding to the pit bottom point set and the pit slope point set are recorded for tracing in the subsequent morphological map.

[0032] (4.4) In the main scour pit area When generating the valley line skeleton, first determine the main direction of the valley line. Specifically, perform principal component analysis on the representative points of the pit edge and take the direction corresponding to the largest eigenvalue as the main direction of the valley line. When the number of representative points at the edge of the pit is less than a preset threshold, the least squares straight line fitting direction of the valley candidate points is used to replace the main direction; the preset threshold is 8 to 12; the source of the main direction of the valley is recorded.

[0033] (4.5) Preset the strip width and divide the main scour pit area along the main direction of the valley line. Divided into several strips, for the main scour pit area For any point within the valley, its projected coordinates along the principal direction of the valley line. for: , in, Here are the plane coordinates of this point. The plane coordinates of the deepest anchor point. The main direction of the valley line; Within each zone, the point with the greatest scour depth is selected as the representative point of the valley line. : , in, For the first If the number of points in a certain strip is less than the preset strip point count threshold, the strip width is increased or the strip is skipped, and the number of points represented by the valley line, the number of valid strips, and the number of missing strips are recorded. The preset strip point count threshold is determined based on the point cloud density of the main scour pit area, and is generally 3 to 10.

[0034] (4.6) Connect the representative points of each valley line in ascending order of projected coordinates to form a valley line skeleton; the valley line connection relationship between adjacent valley line representative points is used to represent the order relationship of the valley line skeleton, and record the spatial distance, strip number and support point index between adjacent valley line representative points; after the valley line skeleton is generated, output the valley line representative point sequence, valley line connection relationship, valley line main direction and valley line quality information; among which, the valley line quality information includes the source of the valley line main direction, the number of effective strips, the number of missing strips and the number of valley line representative points.

[0035] (4.7) After the above processing, the morphological semantic unit is output. The morphological semantic unit includes the pit edge ring, the pit bottom point set, the pit slope point set and the valley line skeleton. The deepest point anchor point is used as the unified positioning reference of the morphological semantic unit and as the source of the measurement feature node in step (5). At the same time, the support point index, generation parameters and quality information corresponding to the morphological semantic unit are output. The generation parameters include the azimuth box step size, the angle gap threshold, the spatial distance gap threshold, the pit bottom-pit slope boundary parameters, the strip width and the strip point number threshold. The quality information includes the pit edge quality information, the pit bottom connectivity screening results and the valley line quality information. The morphological semantic unit is used in step (5) to generate morphological semantic unit nodes, topological point nodes, sequential relation edges, adjacency relation edges and source information.

[0036] (5) The morphological semantic units obtained in step (4) and the measurement feature set obtained in step (3) are used to construct a bridge scour morphology map. The bridge scour morphology map includes a node set, an edge set, node attributes, edge attributes, source information, quality information and graph-level metadata.

[0037] like Figure 5 As shown, step (5) specifically includes the following steps: (5.1) Based on the main scour pit area A node set is constructed from morphological semantic units and measurement feature sets. The node set includes scour pit object nodes, morphological semantic unit nodes, topological point nodes, and measurement feature nodes. Scour pit object nodes refer to the nodes formed by the main scour pit area. The generated corresponding nodes are used to represent a main scour pit as a whole; Morphological semantic unit nodes refer to four types of corresponding nodes generated from the pit bottom point set, pit slope point set, pit edge loop, and valley line skeleton, respectively. Topological point nodes refer to the two corresponding nodes generated by the pit edge representative point and the valley line representative point, respectively. The measurement feature nodes refer to the four corresponding nodes generated by the deepest point anchor point, projected surface area, scour pit volume, and maximum slope. Each node records the node type, geometric location, and corresponding support point index or support grid cell index; the support point index is used to record the original point number of the riverbed point corresponding to the node in the underwater sonar point cloud data.

[0038] (5.2) For each morphological semantic unit node, its geometric position is determined according to the point set corresponding to the node, and the centroid of the support point set is used as the geometric position of the node. : , in, Represents a node. Represents a node The corresponding support point set, Indicates the riverbed points in the support point set; For nodes generated from a point set, the node depth attribute is also determined based on the scour depth values ​​of each riverbed point in the support point set. Specifically, the median of the scour depth values ​​of each riverbed point in the support point set is used as the node depth attribute. : , in, Indicates riverbed point The corresponding scour depth value; For nodes generated from measurement features, the corresponding measurement features are directly written into the node attributes; among them, the deepest point anchor node records the location of the deepest point and the maximum scour depth, the projected surface area node records the projected surface area of ​​the scour pit, the volume node records the volume of the scour pit, and the maximum slope node records the maximum slope value, the location of the maximum slope, the slope direction, and the slope azimuth.

[0039] (5.3) Construct an edge set based on the hierarchical, subordinate, sequential, and adjacency relationships between nodes; Hierarchical relationship edges are used to represent the inclusion relationship between scour pit object nodes and morphological semantic unit nodes; for each morphological semantic unit node, a hierarchical relationship edge is established between it and the scour pit object node. Subordination edges are used to indicate that measurement feature nodes belong to the same main scour pit object; for the deepest point anchor node, projected surface area node, volume node and maximum slope node, respectively establish subordination edges between them and the scour pit object node; Sequential relationship edges are used to represent the arrangement order between topological point nodes; for pit edge topological point nodes, sequential relationship edges are established in order of azimuth angle from small to large; for valley line topological point nodes, sequential relationship edges are established in order of projected coordinates in the main direction of the valley line. Adjacency edges are used to represent the spatial adjacency relationships between morphological semantic units; an adjacency edge is established when the nearest distance between the support point sets corresponding to two morphological semantic units is not greater than a preset adjacency distance threshold. , in, and These are two morphological semantic unit nodes. For morphological semantic unit nodes The corresponding support point set, To support point sets points within, For morphological semantic unit nodes The corresponding support point set, To support point sets points within; To preset the adjacency distance threshold, it is determined based on the average point spacing or grid resolution of the riverbed point set, typically taken as a fraction of the average point spacing or the diagonal length of the grid cell. times.

[0040] (5.4) Connect the topological point nodes generated by the representative points of the pit edge in azimuth order to form the topological structure of the pit edge; when the sequence of representative points of the pit edge satisfies the closure criterion in step (4), establish the first and last closed edges, and record the pit edge closure state, maximum angle gap, maximum spatial gap, angle gap threshold and spatial distance gap threshold in the edge attributes. The topological point nodes generated for the valley line representative points are connected in order of projected coordinates along the main direction of the valley line to form the valley line skeleton topological structure; the strip number, spatial distance and support point index of adjacent valley line representative points are recorded in the edge attributes.

[0041] (5.5) Write source information for nodes and edges. Node source information includes node source, supporting point index, supporting grid cell index and generation parameters; edge source information includes edge type, edge generation basis, related supporting point pairs and related topological point sequence. Among them, the source information of the morphological semantic unit node comes from the support point index obtained in step (4); the source information of the corresponding node generated by the projected surface area and the corresponding node generated by the scour pit volume comes from the grid unit index participating in the area and volume calculation; the source information of the corresponding node generated by the maximum slope comes from the maximum slope location and its neighboring point index or the neighboring grid unit index. By tracing the source information, a correspondence is established from bridge scour morphology maps to riverbed point sets, and then to underwater sonar point cloud data.

[0042] (5.6) Record the number of nodes and the number of edges as quality information, and write quality information into the bridge scour morphology map. The quality information includes the quality information of the pit edge, the screening results of the pit bottom connectivity, the quality information of the valley line, the number of nodes and the number of edges. When the pit edge is not closed, the pit bottom connectivity is insufficient, the number of effective valley strips is less than the preset threshold, or the nodes and edges do not meet the consistency verification requirements, the corresponding abnormal marker is recorded in the quality information.

[0043] (5.7) Combine the node set, edge set, node attributes, edge attributes, traceability information and quality information to form a bridge scour morphology map. The bridge scour morphology map meets the preset map data mode. The map-level metadata includes map identifier, coordinate system, unit, center of main scour pit, node type, edge type, number of nodes, number of edges, upstream data source and processing parameters. Before output, a consistency check is performed on the bridge scour morphology map. The consistency check includes checking the uniqueness of node identifiers, the existence of edge endpoints, the legality of node types, and the legality of edge types. After passing the consistency check, the bridge scour morphology map is output.

[0044] Example 2: A bridge scour morphology map construction system based on underwater sonar point clouds according to the present invention includes: The preprocessed point cloud acquisition module is used to acquire underwater sonar point cloud data within a preset sampling range, with the center of the bridge pier as the reference. For the underwater sonar point cloud data Preprocessing is performed to obtain a preprocessed point cloud. .

[0045] The riverbed point set acquisition module is used to process the preprocessed point cloud. Semantic pre-segmentation is performed to obtain the riverbed point set. and establish riverbed point set underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number.

[0046] The module for calculating measurement features is used to calculate the features based on the riverbed point set. Determine a reference bed model, and calculate the riverbed point set based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area The measurement feature set is calculated, which includes the deepest anchor point, the maximum scour depth, the sequence of representative points on the pit edge, the pit edge depth level, area and volume measurement information, and the maximum slope measurement information.

[0047] The morphological semantic unit module is used to generate morphological semantic units in the main scour pit area. Within the system, based on the non-negative scour depth field, the sequence of representative points on the crater edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the crater edge ring, the crater bottom point set, the crater slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded.

[0048] The map construction module is used to construct maps based on the main scour pit area. The bridge scour morphology map is constructed by measuring feature sets and morphological semantic units. The bridge scour morphology map includes node sets, edge sets, node attributes, edge attributes, source information, quality information and graph-level metadata.

[0049] Example 3: An electronic device according to the present invention includes a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the bridge scour morphology map construction method based on underwater sonar point clouds as described above.

[0050] Example 4: A computer-readable storage medium according to the present invention stores a computer program, which, when executed by a processor, implements the steps of a method for constructing a bridge scour morphology map based on underwater sonar point clouds as described above.

[0051] Example 5: A computer program product of the present invention includes a computer program that, when executed by a processor, implements the steps of a method for constructing a bridge scour morphology map based on underwater sonar point clouds as described above.

Claims

1. A method for constructing bridge scour morphology maps based on underwater sonar point clouds, characterized in that, Includes the following steps: (1) Using the center of the bridge pier as a reference, acquire underwater sonar point cloud data within a preset sampling range. For the underwater sonar point cloud data Preprocessing is performed to obtain a preprocessed point cloud. ; (2) The preprocessed point cloud Semantic pre-segmentation is performed to obtain the riverbed point set. and establish riverbed point set underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number; (3) Based on riverbed point set Determine a reference bed model, and calculate the riverbed point set based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area Calculate the measurement feature set, which includes the deepest anchor point, maximum scour depth, sequence of representative points on the pit edge, pit edge depth level, area and volume measurement information, and maximum slope measurement information. (4) In the main scour pit area Inside, based on the non-negative scour depth field, the sequence of representative points on the pit edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the pit edge ring, the pit bottom point set, the pit slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded. Step (4) specifically includes the following steps: (4.1) Based on the sequence of representative pit edge points obtained in step (3), connect adjacent representative pit edge points in ascending order of azimuth angle to form a candidate pit edge loop; for adjacent representative pit edge points and Calculate the azimuth notch and spatial distance : ， ， in, The sequence index is used to represent the pit edge points; when determining the first and last adjacent points, the last pit edge point in the sequence is connected to the first pit edge point in the sequence, and the azimuth is then used. Periodic compensation; When the adjacent pit edge representative points satisfy ,and At that time, establish the connectivity between representative points on adjacent pit edges; among which, The angle gap threshold is determined based on the azimuth box step size. The spatial distance gap threshold is determined based on the average spacing of the point cloud or the grid resolution. Calculate the maximum angular gap and maximum spatial gap of the candidate ring at the pit edge: ， ， when and When the crater edge representative point sequence is closed into a crater edge loop, it is otherwise retained as an unclosed crater edge. The crater edge closure status, the maximum gap and its corresponding azimuth angle interval are recorded as crater edge quality information. The maximum gap includes the maximum angular gap and the maximum spatial gap. (4.2) Using the deepest anchor point as the radial analysis center, construct the scour depth as a function of radius within each azimuth angle box. A changing radial profile refers to a one-dimensional function in which the scour depth of a riverbed point varies with its planar distance to the deepest anchor point within the same azimuth box. For each radial profile, calculate the first and second order variations of the scour depth, and then determine the boundary radius between the pit bottom and the pit slope based on the location of the abrupt change in depth. Specifically, this involves constructing a fracture scoring function: ， in, For the first Radial profile within an azimuth angle sub-box For first-order changes, It is a second-order change. For curvature surrogate quantity, symbol " "Indicates normalization processing, , , These are the preset weighting coefficients for the fracture scoring function; The location where the fracture score function reaches its maximum value is taken as the pit bottom-pit slope boundary radius in that azimuth direction. : ， Record the pit bottom-pit slope boundary radius and the fracture scoring function weight coefficient in each azimuth direction as the pit bottom-pit slope boundary parameters; (4.3) For the main scour pit area For any point within the deepest point, calculate its radius relative to the deepest anchor point. and azimuth ;when When, the point is assigned to the pit bottom point set; when At that time, the point is assigned to the pit slope point set, where The radius of the pit edge at the corresponding azimuth angle ; Perform connectivity filtering on the pit bottom point set, and retain the connected component with the most points or the largest area as the final pit bottom point set; the remaining scattered points are not considered as the main body of the pit bottom, and the pit slope point set consists of the points between the pit bottom boundary and the pit edge; record the pit bottom connectivity filtering results and the support point indexes corresponding to the pit bottom point set and the pit slope point set. (4.4) In the main scour pit area When generating the valley line skeleton, first determine the main direction of the valley line. Specifically, perform principal component analysis on the representative points of the pit edge and take the direction corresponding to the largest eigenvalue as the main direction of the valley line. When the number of representative points at the edge of the pit is less than the preset threshold, the least squares straight line fitting direction of the valley line candidate points is used to replace the main direction; the source of the main direction of the valley line is recorded. (4.5) Preset the strip width and divide the main scour pit area along the main direction of the valley line. Divided into several strips, for the main scour pit area For any point within the valley, its projected coordinates along the principal direction of the valley line. for: ， in, Let be the plane coordinates of this point. The plane coordinates of the deepest anchor point. The main direction of the valley line; Within each zone, the point with the greatest scour depth is selected as the representative point of the valley line. : ， in, For the first If the number of points in a certain strip is less than the preset threshold for the number of points in a strip, the strip width is increased or the strip is skipped, and the number of points represented by the valley line, the number of valid strips, and the number of missing strips are recorded. (4.6) Connect the representative points of each valley line in ascending order of projected coordinates to form a valley line skeleton; the valley line connection relationship between adjacent valley line representative points is used to represent the order relationship of the valley line skeleton, and record the spatial distance, strip number and support point index between adjacent valley line representative points; after the valley line skeleton is generated, output the valley line representative point sequence, valley line connection relationship, valley line main direction and valley line quality information; among which, the valley line quality information includes the source of the valley line main direction, the number of effective strips, the number of missing strips and the number of valley line representative points; (4.7) Output morphological semantic units, which include pit edge rings, pit bottom point sets, pit slope point sets, and valley line skeletons. The deepest point anchor point serves as the unified positioning reference for the morphological semantic units. At the same time, output the support point index, generation parameters, and quality information corresponding to the morphological semantic units. The generation parameters include azimuth box step size, angle gap threshold, spatial distance gap threshold, pit bottom-pit slope boundary parameters, strip width, and strip point number threshold. The quality information includes pit edge quality information, pit bottom connectivity screening results, and valley line quality information. (5) Based on the main scour pit area The bridge scour morphology map is constructed by measuring feature sets and morphological semantic units. The bridge scour morphology map includes node sets, edge sets, node attributes, edge attributes, source information, quality information and graph-level metadata.

2. The method for constructing bridge scour morphology maps based on underwater sonar point clouds according to claim 1, characterized in that, The preset sampling range in step (1) is determined by the preset sampling radius. Confirmed; the underwater sonar point cloud data This includes the three-dimensional coordinate information of each point. When the sonar device outputs echo intensity information and sampling time information, the echo intensity information and sampling time information are recorded as point attributes; this is the underwater sonar point cloud data. The preprocessing steps of outlier removal, statistical denoising, voxel downsampling, coordinate unification, and normalization are performed sequentially to obtain the preprocessed point cloud. It generates preprocessed metadata including coordinate system, units, normalization method, normalization parameters, statistical denoising parameters, and voxel downsampling parameters.

3. The method for constructing bridge scour morphology maps based on underwater sonar point clouds according to claim 1, characterized in that, In step (2), the preprocessed point cloud Semantic pre-segmentation is performed, and the preprocessed point cloud is segmented based on point cloud normals, local flatness, elevation distribution, and spatial connectivity features. Divided into riverbed point sets and bridge foundation point set Among them, riverbed point set The set of points belonging to the riverbed surface, and the set of bridge foundation points. It is the set of points that belong to the surface of the pier or bridge foundation entity.

4. The method for constructing bridge scour morphology maps based on underwater sonar point clouds according to claim 1, characterized in that, Step (3) specifically includes the following steps: (3.1) For the riverbed point set Robust fitting is performed to obtain a reference bed model, which serves as the riverbed elevation benchmark for calculating scour depth. The reference bed model adopts a local planar model, and its expression is: ， in, , , For reference bed surface parameters, , Here are the plane coordinates of the riverbed point. Predict the elevation of the reference bed surface corresponding to the planar position; at the same time, record the proportion of fitted in-points, the statistics of the fitted residuals and the parameters of the reference bed surface to form the reference bed surface fitting information. (3.2) For the riverbed point set any riverbed point The scour depth of any riverbed point relative to the reference bed surface is calculated based on the reference bed surface model. The calculation formula is: ， in, Riverbed point The measured elevation; when the calculated scour depth value is less than zero, the scour depth value of the riverbed point is counted as zero; from the riverbed point set The non-negative scour depth field is composed of the non-negative scour depth values ​​corresponding to each riverbed point. Then, based on the non-negative scour depth field, the riverbed point set is... Scour candidate points are determined for each riverbed point, and a scour candidate point set is extracted. Specifically, using riverbed points For the target riverbed point, its neighborhood is determined. The riverbed point neighborhood is the set of neighboring points centered on the target riverbed point, determined by a search distance parameter. Based on the elevation difference or scour depth difference between the target riverbed point and its neighboring riverbed points, the elevation undulation characteristics of the target riverbed point are obtained. Then, a local ternary model algorithm is used to encode the elevation undulation characteristics to obtain the scour confidence value of the target riverbed point. Riverbed points with scour confidence values ​​not less than a preset confidence threshold are classified as scour candidate points, and riverbed points with scour confidence values ​​less than the preset confidence threshold are classified as non-scour points. All scour candidate points form a scour candidate point set. ; (3.3) Set of candidate scour points Determining the main scour pit area using a depth threshold method Specifically, it refers to: using the scouring candidate point set The scour depth values ​​constitute a scour depth sample set, and the high quantile of the scour depth sample set is calculated. , median and median absolute deviation and determine the depth threshold. : ， in, To preset the high quantile parameters, To achieve the lower threshold of the scour depth engineering, This is the relative threshold coefficient. These are robust statistical coefficients; scour depth value Not less than the depth threshold Points with the deepest points are selected as depth candidates, and spatial clustering is performed based on the planar coordinates of these candidates to obtain several candidate clusters. Among these clusters, the cluster containing the deepest point is selected as the main scour pit area. ; (3.4) Main scour pit area The point on the riverbed with the largest scour depth is selected as the deepest point anchor point, thus obtaining the location of the deepest point and the maximum scour depth. satisfy: ， in, Indicates riverbed point The scour depth value, (3.5) Using the deepest anchor point as the origin of the polar coordinates, the main scour pit area is... Calculate the polar coordinates of any point on the riverbed: ， ， in, The coordinates of the plane at the origin of the polar coordinate system are... The distance from the riverbed point to the origin of the polar coordinate system is the planar distance. For the corresponding azimuth angle; According to the preset azimuth angle, the box is divided into steps. The azimuth angle is divided into several azimuth bins. Within each azimuth bin, riverbed points with scour depth values ​​not exceeding the candidate depth threshold for the pit edge are selected as candidate pit edge points. The candidate depth threshold for the pit edge is determined by the maximum scour depth and a preset proportional coefficient. Subsequently, the pit edge radius of that azimuth bin is determined by the high quantile of the radius of the candidate pit edge points. : ， in, For the first The set of candidate points for the pit edge within each azimuth angle bin. For quantile operators, The high resolution parameters are preset; then, within this azimuth caliber, the radius of the pit edge is selected. The closest point is used as the representative point of the pit edge, forming a sequence of representative points of the pit edge ordered by azimuth. Pit edge depth horizontal The average value of the scour depth at the representative point on the edge of the pit is taken, based on the statistical values ​​of the scour depth at the representative point on the edge of the pit. (3.6) The non-negative scour depth field is represented by a grid according to the preset grid resolution to obtain the gridded scour depth field, with the scour depth not less than the pit edge depth. The grid cells are used as the projection area of ​​the scour pit, and the projected surface area and volume of the scour pit are calculated. Let the grid cell depth be The area of ​​the grid cell is The projected surface area of ​​the scour pit satisfy: ， in, This is an indicator function; it takes the value 1 if the condition inside the parentheses is true, and 0 otherwise. Based on the depth level of the pit edge Calculate the volume of the scour pit : ， After the calculation is completed, area and volume measurement information is generated; the area and volume measurement information includes projected surface area, scour pit volume, grid resolution, pit edge depth level, and grid cell indexes involved in the calculation; (3.7) Calculate the maximum slope, the location of the maximum slope, and the direction of the slope within the projected area of ​​the scour pit. The local bed elevation is expressed as: ， in, To obtain the slope value, the gradient of the local bed elevation is calculated for the scour depth value at the corresponding planar location: ， Within the projected area of ​​the scour pit, select the point or grid cell with the largest slope value as the location of the maximum slope, and take the slope value corresponding to this location as the maximum slope; the slope direction is taken as the direction of the steepest descent of the bed surface, and its direction vector is: ， Simultaneously, the azimuth angle of the slope direction is recorded. After the calculation is completed, the maximum slope measurement information is generated. The maximum slope measurement information includes the maximum slope value, the location of the maximum slope, the slope direction, the slope azimuth angle, and the neighboring point index or grid cell index that participated in the slope calculation.

5. The method for constructing bridge scour morphology maps based on underwater sonar point clouds according to claim 4, characterized in that, Step (5) specifically includes the following steps: (5.1) Based on the main scour pit area A node set is constructed from morphological semantic units and measurement feature sets. The node set includes scour pit object nodes, morphological semantic unit nodes, topological point nodes, and measurement feature nodes. Scour pit object nodes refer to the nodes formed by the main scour pit area. The generated corresponding nodes are used to represent a main scour pit as a whole; Morphological semantic unit nodes refer to four types of corresponding nodes generated from the pit bottom point set, pit slope point set, pit edge loop, and valley line skeleton, respectively. Topological point nodes refer to the two corresponding nodes generated by the pit edge representative point and the valley line representative point, respectively. The measurement feature nodes refer to the four corresponding nodes generated by the deepest point anchor point, projected surface area, scour pit volume, and maximum slope. Each node records the node type, geometric location, and corresponding support point index or support grid cell index; the support point index is used to record the original point number of the riverbed point corresponding to the node in the underwater sonar point cloud data; (5.2) For each morphological semantic unit node, its geometric position is determined according to the point set corresponding to the node, and the centroid of the support point set is used as the geometric position of the node. : ， in, Represents a node. Represents a node The corresponding support point set, Indicates the riverbed points in the support point set; For nodes generated from a point set, the node depth attribute is also determined based on the scour depth values ​​of each riverbed point in the support point set. Specifically, the median of the scour depth values ​​of each riverbed point in the support point set is used as the node depth attribute. : ， in, Indicates riverbed point The corresponding scour depth value is used to write the node's geometric position and node depth attributes into the node attributes. For nodes generated from measurement features, the corresponding measurement features are directly written into the node attributes; among them, the corresponding node generated from the deepest point anchor point records the location of the deepest point and the maximum scour depth, the corresponding node generated from the projected surface area records the projected surface area of ​​the scour pit, the corresponding node generated from the scour pit volume records the volume of the scour pit, and the corresponding node generated from the maximum slope records the maximum slope value, the location of the maximum slope, the slope direction, and the slope azimuth. (5.3) Construct an edge set based on the hierarchical, subordinate, sequential, and adjacency relationships between nodes; Hierarchical relationship edges are used to represent the inclusion relationship between scour pit object nodes and morphological semantic unit nodes; for each morphological semantic unit node, a hierarchical relationship edge is established between it and the scour pit object node. Subordination edges are used to indicate that measurement feature nodes belong to the same main scour pit object; for the deepest point anchor node, projected surface area node, volume node and maximum slope node, respectively establish subordination edges between them and the scour pit object node; Sequential relationship edges are used to represent the arrangement order between topological point nodes; for pit edge topological point nodes, sequential relationship edges are established in order of azimuth angle from small to large; for valley line topological point nodes, sequential relationship edges are established in order of projected coordinates in the main direction of the valley line. Adjacency edges are used to represent the spatial adjacency relationships between morphological semantic units; an adjacency edge is established when the nearest distance between the support point sets corresponding to two morphological semantic units is not greater than a preset adjacency distance threshold. ， in, and These are two morphological semantic unit nodes. For morphological semantic unit nodes The corresponding support point set, To support point sets points within, For morphological semantic unit nodes The corresponding support point set, To support point sets points within; The preset adjacency distance threshold is determined based on the average point spacing or grid resolution of the riverbed point set; (5.4) Connect the topological point nodes generated from the representative points of the pit edge in azimuth order to form the pit edge topological structure; when the sequence of representative points of the pit edge satisfies and Establish closed edges at both ends, and record the pit edge closure status, maximum angle gap, maximum spatial gap, angle gap threshold, and spatial distance gap threshold in the edge attributes; The topological point nodes generated for the valley line representative points are connected in the order of their projected coordinates along the main direction of the valley line to form the valley line skeleton topological structure; the strip number, spatial distance, and support point index of adjacent valley line representative points are recorded in the edge attributes. (5.5) Write source information for nodes and edges. Node source information includes node source, supporting point index, supporting grid cell index and generation parameters; edge source information includes edge type, edge generation basis, related supporting point pairs and related topological point sequence. Among them, the source information of the morphological semantic unit node comes from the support point index; the source information of the corresponding node generated by the projected surface area and the corresponding node generated by the scour pit volume comes from the grid cell index participating in the area and volume calculation; the source information of the corresponding node generated by the maximum slope comes from the maximum slope location and its neighboring point index or neighboring grid cell index. Establish a correspondence between bridge scour morphology maps, riverbed point sets, and underwater sonar point cloud data by tracing the source information; (5.6) Record the number of nodes and the number of edges as quality information, and write quality information into the bridge scour morphology map. The quality information includes the quality information of the pit edge, the screening results of the pit bottom connectivity, the quality information of the valley line, the number of nodes and the number of edges. When the pit edge is not closed, the pit bottom connectivity is insufficient, the number of effective valley strips is less than the preset threshold, or the nodes and edges do not meet the consistency verification requirements, the corresponding abnormal marker is recorded in the quality information. (5.7) Combine the node set, edge set, node attributes, edge attributes, traceability information and quality information to form a bridge scour morphology map. The bridge scour morphology map meets the preset map data mode. The map-level metadata includes map identifier, coordinate system, unit, center of main scour pit, node type, edge type, number of nodes, number of edges, upstream data source and processing parameters. Before output, a consistency check is performed on the bridge scour morphology map. The consistency check includes checking the uniqueness of node identifiers, the existence of edge endpoints, the legality of node types, and the legality of edge types. After passing the consistency check, the bridge scour morphology map is output.

6. A system for constructing bridge scour morphology maps based on underwater sonar point clouds, characterized in that, include: The preprocessed point cloud acquisition module is used to acquire underwater sonar point cloud data within a preset sampling range, with the center of the bridge pier as the reference. For the underwater sonar point cloud data Preprocessing is performed to obtain a preprocessed point cloud. ; The riverbed point set acquisition module is used to process the preprocessed point cloud. Semantic pre-segmentation is performed to obtain the riverbed point set. and establish riverbed point set underwater sonar point cloud data Point cloud index mapping between points, used to record riverbed point sets. Underwater sonar point cloud data for various points The corresponding original point number; The calculation and measurement feature module is used to calculate the riverbed point set. Determine a reference bed model, and calculate the riverbed point set based on the reference bed model. The scour depth values ​​of each point relative to the reference bed surface are calculated, and scour depth values ​​less than zero are counted as zero to obtain a non-negative scour depth field; then, based on the non-negative scour depth field, the data are obtained from the riverbed point set. Extracting the candidate set of scour points and from the scouring candidate point set Determining the main scour pit area Meanwhile, based on the main scour pit area Calculate the measurement feature set, which includes the deepest anchor point, maximum scour depth, sequence of representative points on the pit edge, pit edge depth level, area and volume measurement information, and maximum slope measurement information. The morphological semantic unit module is used to generate morphological semantic units in the main scour pit area. Inside, based on the non-negative scour depth field, the sequence of representative points on the pit edge, and the deepest point anchor, morphological semantic units are generated. The morphological semantic units include the pit edge ring, the pit bottom point set, the pit slope point set, and the valley line skeleton. The deepest point anchor serves as the unified positioning reference for the morphological semantic units, and the support point index, generation parameters, and quality information corresponding to the morphological semantic units are recorded. In the morphological semantic unit generation module, based on the obtained sequence of pit edge representative points, adjacent pit edge representative points are connected in ascending order of azimuth angle to form a pit edge candidate loop; for adjacent pit edge representative points... and Calculate the azimuth notch and spatial distance : ， ， in, The sequence index is used to represent the pit edge points; when determining the first and last adjacent points, the last pit edge point in the sequence is connected to the first pit edge point in the sequence, and the azimuth is then used. Periodic compensation; When the adjacent pit edge representative points satisfy ,and At that time, establish the connectivity between representative points on adjacent pit edges; among which, The angle gap threshold is determined based on the azimuth box step size. The spatial distance gap threshold is determined based on the average spacing of the point cloud or the grid resolution. Calculate the maximum angular gap and maximum spatial gap of the candidate ring at the pit edge: ， ， when and When the crater edge representative point sequence is closed into a crater edge loop, it is otherwise retained as an unclosed crater edge. The crater edge closure status, the maximum gap and its corresponding azimuth angle interval are recorded as crater edge quality information. The maximum gap includes the maximum angular gap and the maximum spatial gap. Using the deepest anchor point as the radial analysis center, the scour depth is constructed as a function of radius within each azimuth box. A changing radial profile refers to a one-dimensional function in which the scour depth of a riverbed point varies with its planar distance to the deepest anchor point within the same azimuth box. For each radial profile, calculate the first and second order variations of the scour depth, and then determine the boundary radius between the pit bottom and the pit slope based on the location of the abrupt change in depth. Specifically, this involves constructing a fracture scoring function: ， in, For the first Radial profile within an azimuth angle sub-box For first-order changes, It is a second-order change. For curvature surrogate quantity, symbol " "Indicates normalization processing, , , These are the preset weighting coefficients for the fracture scoring function; The location where the fracture score function reaches its maximum value is taken as the pit bottom-pit slope boundary radius in that azimuth direction. : ， Record the pit bottom-pit slope boundary radius and the fracture scoring function weight coefficient in each azimuth direction as the pit bottom-pit slope boundary parameters; For the main scour pit area For any point within the deepest point, calculate its radius relative to the deepest anchor point. and azimuth ;when When, the point is assigned to the pit bottom point set; when At that time, the point is assigned to the pit slope point set, where The radius of the pit edge at the corresponding azimuth angle ; Perform connectivity filtering on the pit bottom point set, and retain the connected component with the most points or the largest area as the final pit bottom point set; the remaining scattered points are not considered as the main body of the pit bottom, and the pit slope point set consists of the points between the pit bottom boundary and the pit edge; record the pit bottom connectivity filtering results and the support point indexes corresponding to the pit bottom point set and the pit slope point set. In the main scour pit area When generating the valley line skeleton, first determine the main direction of the valley line. Specifically, perform principal component analysis on the representative points of the pit edge and take the direction corresponding to the largest eigenvalue as the main direction of the valley line. When the number of representative points at the edge of the pit is less than the preset threshold, the least squares straight line fitting direction of the valley line candidate points is used to replace the main direction; the source of the main direction of the valley line is recorded. The preset strip width is used to define the main scour pit area along the main direction of the valley line. Divided into several strips, for the main scour pit area For any point within the valley, its projected coordinates along the principal direction of the valley line. for: ， in, Let be the plane coordinates of this point. The plane coordinates of the deepest anchor point. The main direction of the valley line; Within each zone, the point with the greatest scour depth is selected as the representative point of the valley line. : ， in, For the first If the number of points in a certain strip is less than the preset threshold for the number of points in a strip, the strip width is increased or the strip is skipped, and the number of points represented by the valley line, the number of valid strips, and the number of missing strips are recorded. The representative points of each valley line are connected in ascending order of projected coordinates to form a valley line skeleton. The valley line connection relationship between adjacent representative points is used to represent the order relationship of the valley line skeleton, and the spatial distance, strip number, and support point index between adjacent representative points are recorded. After the valley line skeleton is generated, the valley line representative point sequence, valley line connection relationship, valley line main direction, and valley line quality information are output. Among them, the valley line quality information includes the source of the valley line main direction, the number of valid strips, the number of missing strips, and the number of valley line representative points. The system outputs morphological semantic units, which include pit edge rings, pit bottom point sets, pit slope point sets, and valley line skeletons. The deepest point anchor serves as the unified positioning reference for the morphological semantic units. Simultaneously, it outputs the support point index, generation parameters, and quality information corresponding to each morphological semantic unit. Generation parameters include azimuth bin step size, angle gap threshold, spatial distance gap threshold, pit bottom-pit slope boundary parameters, strip width, and strip point number thresholds. Quality information includes pit edge quality information, pit bottom connectivity filtering results, and valley line quality information. The map construction module is used to construct maps based on the main scour pit area. The bridge scour morphology map is constructed by measuring feature sets and morphological semantic units. The bridge scour morphology map includes node sets, edge sets, node attributes, edge attributes, source information, quality information and graph-level metadata.

7. An electronic device, characterized in that, Including processor and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method for constructing a bridge scour morphology map based on underwater sonar point clouds as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing a bridge scour morphology map based on underwater sonar point clouds as described in any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing a bridge scour morphology map based on underwater sonar point clouds as described in any one of claims 1-5.

Images