Bridge scour curved surface morphological feature reconstruction method based on three-dimensional sonar point clouds

Abstract

Disclosed in the present invention is a bridge scour curved surface morphological feature reconstruction method based on three-dimensional sonar point clouds, comprising the following steps: collecting point cloud data of a three-dimensional morphology of an underwater riverbed terrain near a bridge foundation, and acquiring original point cloud data of a complete morphology of an underwater riverbed; performing denoising and ball pivoting reconstruction preprocessing on the collected original point cloud data to complete preliminary reconstruction of point clouds; on the basis of the point cloud data having undergone preliminary reconstruction, performing point cloud binarization-based scour morphology recognition on the basis of a computer vision principle, calculating scour pit element classification values of all the point clouds, and recognizing scour pit point clouds; on the basis of the recognized scour pit point clouds, using a K-means clustering algorithm for clustering, and calculating the maximum depth of each pier scour pit; and on the basis of the obtained scour pit element classification values of all the point clouds, providing a curved surface reconstruction method based on scour recognition to perform high-precision curved surface reconstruction on the scour pits. The present invention can achieve high-precision scour pit morphology reconstruction.

Description

Bridge scour curved surface form feature reconstruction method based on three-dimensional sonar point cloud TECHNICAL FIELD

[0001] The present application belongs to the technical field of bridge underwater scour form identification, and particularly relates to a bridge scour curved surface form feature reconstruction method based on three-dimensional sonar point cloud. BACKGROUND

[0002] Scour is the process of water erosion of riverbed, embankment and bridge foundation and other structures around the silt and other materials. It mainly manifests as natural evolution scour, general scour and local scour. Bridge piers will block the water flow to form high-speed water flow on both sides of the bridge pier, causing the riverbed around the bridge pier to deform sharply and form a local scour pit. Local scour will erode the soil around the bridge foundation and hollow out the bridge foundation, causing the bearing capacity of the bridge foundation to decrease, and in severe cases, leading to bridge collapse. At present, scour is gradually becoming the dominant cause of bridge collapse worldwide, and more than half of the bridges are disabled due to scour.

[0003] Scour monitoring during the bridge operation period helps to reasonably develop the bridge operation scheme and emergency engineering measures during the flood period, and is the most effective method to reduce the damage of the bridge due to excessive scour of the foundation. In recent years, with the maturity of sonar technology in underwater application, although certain achievements have been made in detecting the underwater structure of the bridge by using sonar technology, the current scour detection based on three-dimensional sonar point cloud mainly relies on experienced technicians to segment the scour pit point cloud of each bridge pier one by one and obtain the maximum scour depth for description and explanation, and there is a lack of research on three-dimensional reconstruction of bridge scour pits. The above status makes the automation, visualization degree and efficiency low, and the three-dimensional sonar point cloud information is not fully utilized.

[0004] The underwater riverbed point cloud obtained based on three-dimensional sonar technology is usually complex in terrain and sparse in point cloud, and has many holes. The challenges are as follows. On the one hand, the riverbed terrain undulates, and the bridge scour pit point cloud across different water areas cannot be segmented by contour lines. On the other hand, due to the limitations of the acoustic principle of the instrument and the interference of underwater organisms, the point cloud obtained by scanning still has holes and is sparse even after removing noise. However, due to the limitations of the complex underwater detection environment, the current research on bridge wide-area bridge scour form identification is lacking, and the conventional point cloud reconstruction method cannot be well applied to the high-precision reconstruction of the large amount of underwater scour pit sonar point cloud which is sparse.

[0005] Therefore, it is urgent to solve the above problems.

[0006] SUMMARY

[0007] Invention purposes: The purpose of the present application is to provide a bridge scour curved surface morphology feature reconstruction method based on three-dimensional sonar point cloud, which can automatically identify, segment and reconstruct the sonar point cloud of wide area scanning, has strong robustness, and can provide accurate data support for subsequent bridge underwater foundation maintenance decision-making.

[0008] Technical scheme: In order to achieve the above purpose, the present application discloses a bridge scour curved surface morphology feature reconstruction method based on three-dimensional sonar point cloud, comprising the following steps:

[0009] (1) Collecting the point cloud data of the three-dimensional morphology of the underwater riverbed topography near the bridge foundation, using a three-dimensional real-time sonar system to carry out multi-directional scanning around the bridge underwater riverbed topography and real-time splicing, so as to obtain the original point cloud data of the complete morphology of the underwater riverbed;

[0010] (2) De-noising and spherical rotation reconstruction preprocessing are performed on the collected original point cloud data to complete the preliminary reconstruction of the point cloud;

[0011] (3) According to the point cloud data after preliminary reconstruction in step (2), the scour morphology recognition of point cloud binary is carried out based on the principle of computer vision, the scour pit element classification value of all point clouds is calculated, and the scour pit point cloud is identified;

[0012] (4) According to the scour pit point cloud identified in step (3), K-mean clustering algorithm is used for clustering and calculating the maximum depth of each pier scour pit;

[0013] (5) According to the scour pit element classification value of all point clouds obtained in step (3), a curved surface reconstruction method based on scour recognition is proposed to carry out high-precision curved surface reconstruction on the scour pit.

[0014] Among them, step (2) specifically comprises the following steps:

[0015] (2.1) After the attitude correction processing and noise interference preprocessing of each measurement data, the measurement lines are merged;

[0016] (2.2) The obtained water depth and position are processed, and the noise interference in the overlapping coverage range of two adjacent measurement lines is removed one by one;

[0017] (2.3) Finally, the point cloud after de-noising and merging is preliminarily reconstructed into a triangular mesh by using the spherical rotation reconstruction method.

[0018] Preferably, in step (2.3), firstly, search is conducted by calculating the upper limit of the density of the sonar point cloud, then a seed triangle is formed by specifying a sphere with radius p touching no other points, and the sphere is rotated around the edge until it touches the third point to form a new triangle, and finally, the preliminary reconstruction of the point cloud is completed by traversing all the reachable edges to form a triangular mesh.

[0019] Further, step (3) specifically comprises the following steps:

[0020] (3.1) Calculate the distance and semi-variance between all point clouds after preliminary reconstruction, and select a Gaussian model as the theoretical variogram function to fit the semi-variance and distance as a function;

[0021] (3.2) Set the search radius L and the skip radius R, and based on the Kriging interpolation, search within the radius range and outside the skip radius in the eight main directions of each point cloud to obtain the interpolated point cloud on the eight elevation profiles;

[0022] (3.3) According to the eight elevation profiles of the interpolated point cloud obtained in step (3.2), calculate the upper opening angle and the lower opening angle of the maximum elevation angle and the minimum elevation angle

[0023] (3.4) Set the flatness t varying with the height of the point cloud, and calculate the elevation identifier symbol of each profile according to the upper and lower opening angles obtained in step (3.3);

[0024] (3.5) According to the principle of local ternary value mode, calculate the (+1) symbol and the (-1) symbol of the elevation identifier symbol on the eight elevation profiles of each point cloud to establish the scour identification search coordinates

[0025] (3.6) According to the scour identification search coordinates obtained in step (3.5), calculate the scour pit element classification value θ of each point cloud;

[0026] (3.7) Calculate the average value θ p of the scour pit element classification value θ of all point clouds as the threshold for segmenting the scour pit point cloud;

[0027] (3.8) For parameters such as search radius, skip radius, and flatness within the range, change them one by one according to the step size and perform identification calculation. When the identification accuracy meets the extraction requirements, perform scour identification on the underwater riverbed point cloud data of the bridge according to the optimized parameters to obtain the scour pit point cloud.

[0028] Further, in step (3.1), the distances between all sample points are calculated and sorted in ascending order. Then, a Gaussian model is used to fit the relationship between the semivariance and the distance. The general formula for the Gaussian model is:

[0029] Where γ represents the semivariance, h represents the distance, C0 is the block constant, C is the camber height, and a is the range, all of which are parameters that need to be preset when using Kriging interpolation.

[0030] Preferably, in step (3.4), the flatness t that varies with height is set as a single straight line model:

[0031] Where t is the flatness, t min To minimize flatness, it is typically set to 0, t max To achieve maximum flatness, it needs to be determined through optimization testing. max z represents the highest elevation of the point cloud to be identified. min The lowest elevation of the point cloud to be identified.

[0032] Furthermore, the calculation value of the scour pit element classification value θ defined in step (3.6) is as follows: First, calculate the scour identification lookup coordinates. The identification symbols representing the eight profiles in a local ternary model. It is the number of (+1) symbols. It is the number of (-1) symbols; then calculate mapping point The distance from the mapping point (8,0) of the pit element is used to define the classification value θ for each point cloud scour pit element, and its calculation formula is as follows:

[0033] Furthermore, the spatial geometric features of the sonar point cloud obtained in step (3.8) include: minimum average spacing and elevation variation coefficient; a randomly generated prediction pilot cloud with the same spatial geometric features as the sonar point cloud, consisting of several Gaussian amplitudes, is used. The area below the horizontal plane is marked as a scour pit point cloud. The prediction pilot cloud is then identified. If the identification result satisfies the recall rate R... e A precision P greater than 0.85, an F1 score greater than 0.89, and a precision P-value greater than 0.85. r Greater than 0.98 and accuracy A cWhen the value reaches 0.93, the scouring recognition is performed on the sonar point cloud scanned according to the tuning parameter, and the evaluation indexes of accuracy A c , recall R e , precision P r and F1 value are calculated according to the following formulas:

[0034] Where TP is true positive, which means that the target point cloud is the point cloud in the scour pit and is marked as the sample point of the scour pit point cloud; FP is false positive, which means that the target point cloud is the non-scour pit point cloud and is identified as the scour pit point cloud; TN is true negative, which means that the target point cloud is the non-scour pit point cloud and is marked as the non-scour pit point cloud; and FN is false negative, which means that the target point cloud is the scour pit point cloud and is marked as the non-scour pit point cloud.

[0035] Further, step (4) specifically comprises the following steps:

[0036] (4.1) According to the scour pit point cloud identified in step (3), the mean and variance of the coordinates of the scour pit point cloud in the transverse direction of the bridge are calculated, and denoising is performed;

[0037] (4.2) Set the number of piers in the range of the sonar point cloud as k, and initialize k samples as the centers a k of the initial clusters;

[0038] (4.3) For each point cloud p in the sonar point cloud, calculate the distance of the point cloud p to the k cluster centers and attribute it to the class corresponding to the cluster center with the smallest distance;

[0039] The distance of the point cloud p to the k cluster centers is calculated by using the squared Euclidean distance and is attributed to the class corresponding to the cluster center with the smallest distance, and the calculation formula of the squared Euclidean distance is as follows: d(x, c) = (x - c) 2

[0040] Where d represents the distance, x represents the spatial coordinate row vector of the point cloud, and c represents the spatial coordinate row vector of the center of the cluster;

[0041] (4.5) Repeat step (4.2) and step (4.3) until the iteration number reaches the upper limit, and obtain the clustering result of the scour pit point cloud of each pier;

[0042] (4.6) For the scour pit point cloud of each cluster of the pier, calculate the minimum elevation to obtain the maximum depth of each scour pit of the sonar scanned bridge.

[0043] Preferably, step (5) specifically comprises the following steps:

[0044] (5.1) Calculate the incenter of each triangle in the triangle mesh obtained from the initial spherical rotation algorithm in step (2);

[0045] (5.2) Calculate the normal vector of each point cloud by covariance analysis of the neighboring points of the point cloud obtained by KNN algorithm;

[0046] Construct the KD-Tree search structure of the point cloud data, search the local neighborhood of each point cloud, and calculate the centroid p of the field c Then calculate the covariance matrix of the point cloud p i as follows: M =∑(p i -p c )(p i -p c ) T

[0047] In the formula, M represents the covariance matrix, and the eigenvalue of the covariance matrix is obtained by singular value decomposition method, wherein the smallest eigenvalue represents the normal vector of the point;

[0048] (5.3) Calculate the tangent plane of each point in the point cloud;

[0049] (5.4) Calculate the projection point of the incenter of the adjacent triangle of each point in the point cloud to the tangent plane of the point, and form the adjustable polygon by the projection point, adjust the size of the adjustable polygon by moving the position of the straight line on the point and the projection point, and the weight of the adjustment is the classification value θ of the scour pit element calculated in step (3);

[0050] (5.5) Update the position of the adjustable triangle, move the vertex of the original triangle mesh to the position of the projection point, and sequentially traverse all the triangles to construct a new adjustable triangle;

[0051] (5.6) According to the principle of the shortest diagonal, divide the space quadrilateral between the adjustable polygon and the adjustable triangle to obtain the triangulated patch;

[0052] (5.7) Integrate the adjustable triangle, the adjustable polygon, and the triangulated patch to obtain the final reconstructed scour pit surface.

[0053] Beneficial effects: Compared with the prior art, the present application has the following remarkable advantages: the present application uses a sonar system to collect point cloud data of the three-dimensional form of the bridge underwater riverbed, based on the principle of computer vision and local ternary value mode, point cloud binary is realized by introducing the classification and identification value of the scour pit, the purpose of efficient and accurate form identification of the bridge underwater wide-area scour pit based on three-dimensional digital model is achieved, K-means clustering is carried out based on the binary identification result, and the maximum scour depth of each pier of the bridge is calculated, which provides technical support for non-contact detection of bridge underwater scour based on three-dimensional sonar; at the same time, the present application increases the number of point clouds to improve the density of the reconstructed target point cloud, solves the problem of sparsity of underwater sonar point cloud, combines the information of the local elevation of the point cloud, and the reconstructed surface has better fitting accuracy and smaller error, which can realize high-precision scour pit form reconstruction, and provide technical support for high-precision quantification of bridge underwater scour based on three-dimensional sonar. BRIEF DESCRIPTION OF DRAWINGS

[0054] Fig. 1 is an implementation flowchart of the present application;

[0055] Fig. 2 is a schematic diagram of the point cloud local ternary value mode used in the present application;

[0056] Fig. 3 is a schematic diagram of the point cloud binary conversion of the present application;

[0057] Fig. 4 is a result diagram of the present application for realizing the form identification of the scour pit point cloud;

[0058] Fig. 5 is a result diagram of the present application for realizing the form reconstruction of the bridge scour pit. DETAILED DESCRIPTION

[0059] The technical solutions of the present application will be further described below in combination with the drawings.

[0060] The present application mainly serves the technical field of bridge underwater scour form identification and reconstruction, and the applied object is the field measurement data of the bridge underwater structure.

[0061] As shown in Fig. 1, a bridge scour surface form feature reconstruction method based on three-dimensional sonar point cloud, comprising the following steps:

[0062] (1) collecting point cloud data of the three-dimensional form of the underwater riverbed terrain near the bridge foundation, using a three-dimensional real-time sonar system to carry out multi-directional scanning and real-time splicing around the underwater riverbed terrain of the bridge by a water measurement ship, so as to obtain original point cloud data of the complete form of the underwater riverbed;

[0063] (2) denoising and spherical rotation reconstruction preprocessing are performed on the original point cloud data collected, and the preliminary reconstruction of the point cloud is completed;

[0064] Step (2) specifically comprises the following steps:

[0065] (2.1) After the attitude correction and noise interference preprocessing of each strip of measured data, the measured lines are merged;

[0066] (2.2) The obtained water depth and position are processed, and the noise interference in the overlapping coverage range of two adjacent measured lines is removed one by one;

[0067] (2.3) Finally, the point cloud after denoising and merging is preliminarily reconstructed by using the spherical rotation reconstruction method to form a triangular mesh;

[0068] First, search by calculating the upper limit of the density of the sonar point cloud, then specify a radius p ball to contact any three points that do not contain other points to form a seed triangle, and rotate the ball around the edge until it touches the third point to form a new triangle, and finally traverse all reachable edges to complete the preliminary reconstruction of the point cloud to form a triangular mesh;

[0069] (3) According to the point cloud data after preliminary reconstruction in step (2), the scour morphology recognition of point cloud binary is carried out based on the principle of computer vision, and the scour pit element classification value of all point clouds is calculated to identify the point cloud of scour pit;

[0070] As shown in FIGS. 2 and 3, the step (3) specifically includes the following steps:

[0071] (3.1) Calculate the distance and semi-variance between all point clouds after preliminary reconstruction, and select a Gaussian model as a theoretical variogram function to fit the semi-variance and distance function;

[0072] Calculate the distance between all sample points and sort them in ascending order, then use a Gaussian model to fit the relationship between semi-variance and distance, and the general formula of the Gaussian model is:

[0073] Where γ represents the semi-variance, h represents the distance, C0 is the block constant, C is the arch height, a is the variable range, and all are parameters that need to be preset when using Kriging interpolation;

[0074] (3.2) Set the search radius l and the skip radius R, and based on Kriging interpolation, find the interpolation point cloud on the eight main direction profiles within the radius range and outside the skip radius at each point cloud;

[0075] (3.3) According to the eight elevation profiles of the interpolation point cloud obtained in step (3.2), calculate the upper opening angle and the lower opening angle of the maximum elevation angle and the minimum elevation angle

[0076] (3.4) Set the flatness t varying with the height of the point cloud, and calculate the elevation identifier symbol of each profile according to the upper and lower opening angles obtained in step (3.3)

[0077] The flatness t varying with the height is set as a single straight line model:

[0078] Where t is the flatness, t min is the minimum flatness, generally set to 0, t max is the maximum flatness, which needs to be determined by tuning tests, z max is the highest elevation of the point cloud to be identified, z min is the lowest elevation of the point cloud to be identified;

[0079] (3.5) According to the principle of local ternary value mode, calculate the (+1) symbol in the elevation identifier symbol of each point cloud on the eight elevation profiles (-1) symbol Establish the scour identification search coordinates

[0080] (3.6) According to the scour identification search coordinates obtained in step (3.5) Calculate the scour pit element classification value θ of each point cloud;

[0081] The calculation value of the scour pit element classification value θ is defined as follows: first, calculate the scour identification search coordinates , which represents the identification symbol of the eight profiles in the local ternary mode, is the number of (+1) symbols, is the number of (-1) symbols; then calculate the mapping point the distance between the mapping point (8, 0) of the pit element, to define the scour pit element classification value θ of each point cloud, and the calculation formula is as follows:

[0082] (3.7) Calculate the average value θ p of the scour pit element classification value θ of all point clouds as the threshold for segmenting the scour pit point cloud;

[0083] (3.8) Change the parameters search radius, skip radius, and flatness one by one in the value range and perform identification calculation according to the step length, and when the identification accuracy meets the extraction requirements, perform scour identification on the bridge underwater riverbed point cloud data according to the optimized parameters, and identify the scour pit point cloud;

[0084] The spatial geometric features of the obtained sonar point cloud by statistical scanning include: minimum average spacing and height variation coefficient; a pre-test point cloud with the same spatial geometric features as the sonar point cloud is randomly generated, and the point cloud below the horizontal plane is marked as an erosion pit point cloud; the pre-test point cloud is identified, and if the identification result meets the recall rate R e greater than 0.85, the F1 value is greater than 0.89, the precision P r is greater than 0.98, and the accuracy A c is greater than 0.93, then the sonar point cloud obtained by scanning is identified according to the erosion according to the optimized parameters; the calculation formula of the above evaluation indexes is:

[0085] wherein TP is true positive, which refers to a sample point of the target point cloud being an erosion pit point cloud and being marked as an erosion pit point cloud; FP is false positive, which refers to a sample point of the target point cloud being a non-erosion pit point cloud and being identified as an erosion pit point cloud; TN is true negative, which refers to a sample point of the target point cloud being a non-erosion pit point cloud and being marked as a non-erosion pit point cloud; and FN is false negative, which refers to a sample point of the target point cloud being an erosion pit point cloud and being marked as a non-erosion pit point cloud;

[0086] (4) According to the erosion pit point cloud identified in step (3), a K-mean clustering algorithm is used for clustering and the maximum depth of each pier erosion pit is calculated;

[0087] As shown in FIG. 4, the step (4) specifically includes the following steps:

[0088] (4.1) According to the erosion pit point cloud identified in step (3), the coordinate mean and variance of the erosion pit point cloud in the transverse bridge direction are calculated, and denoising is performed;

[0089] (4.2) The number k of piers in the range of the sonar point cloud is set, and k samples are initialized as the initial clustering centers a k ;

[0090] (4.3) For each point cloud p in the sonar point cloud, the distance of the point cloud p to the k clustering centers is calculated and the point cloud p is attributed to the class corresponding to the clustering center with the minimum distance;

[0091] The distance of the point cloud p to the k clustering centers is calculated by using the squared Euclidean distance, and the point cloud p is attributed to the class corresponding to the clustering center with the minimum distance, and the calculation formula of the squared Euclidean distance is as follows: d(x, c) = (x - c) 2

[0092] wherein d represents the distance, x represents the spatial coordinate row vector of the point cloud, and c represents the spatial coordinate row vector of the center of the cluster;

[0093] (4.5) Repeat step (4.2) and step (4.3) until the iteration number reaches the upper limit, and obtain the clustering results of each pier scour pit point cloud;

[0094] (4.6) For each cluster of pier scour pit point cloud, calculate the minimum value of its height to obtain the maximum depth of each scour pit of the sonar scanned bridge;

[0095] (5) According to the scour pit element classification value of all point clouds obtained in step (3), a curved surface reconstruction method based on scour identification is proposed to reconstruct the scour pit with high precision;

[0096] As shown in Figure 5, the step (5) specifically includes the following steps:

[0097] (5.1) According to the three-dimensional mesh obtained by the preliminary spherical rotation algorithm in step (2), calculate the incenter of each triangle in the mesh;

[0098] (5.2) According to the KNN algorithm, obtain the adjacent points of the point cloud to perform covariance analysis to estimate the normal vector of each point cloud;

[0099] Construct a KD-Tree search structure of the scanned point cloud data, search the local neighborhood of each point cloud, and calculate the centroid p of the field c , then calculate the covariance matrix of the point cloud p i as follows: M = Σ(p i -p c )(p i -p c ) T

[0100] In the formula, M represents the covariance matrix, and the eigenvalue of the covariance matrix is obtained according to the singular value decomposition method, wherein the smallest eigenvalue represents the normal vector of the point;

[0101] (5.3) Calculate the tangent plane of each point in the point cloud;

[0102] (5.4) Calculate the projection point of the incenter of the adjacent triangle of each point in the point cloud to the tangent plane of the point, and form an adjustable polygon by the projection point. Adjust the size of the adjustable polygon by moving the position of the straight line on the point and the projection point, and the weight of the adjustment is the scour pit element classification value θ calculated in step (3.6);

[0103] (5.5) Update the adjustable triangle position, move the vertex of the original triangular mesh to the position of the projection point, and sequentially traverse all triangles to construct a new adjustable triangle;

[0104] (5.6) Triangulate the space quadrilateral between the adjustable polygon and the adjustable triangle by the principle of the shortest diagonal to get the triangulated patch;

[0105] (5.7) Integrate the adjustable triangle, the adjustable polygon and the triangulated patch to get the final reconstructed scour pit surface.

Claims

1. A method for reconstructing morphological characteristics of a bridge scour surface based on a three-dimensional sonar point cloud, characterized in that, It comprises the following steps: (1) Collecting the point cloud data of the three-dimensional form of the underwater riverbed terrain near the bridge foundation, using a three-dimensional real-time sonar system to carry out multi-directional scanning around the bridge underwater riverbed terrain and real-time splicing, thereby obtaining the original point cloud data of the complete form of the underwater riverbed; (2) Preprocessing the original point cloud data obtained by denoising and spherical rotation reconstruction to complete the preliminary reconstruction of the point cloud; (3) According to the preliminary reconstructed point cloud data of step (2), performing erosion form identification of point cloud binarization based on the principle of computer vision, calculating the erosion pit element classification value of all point clouds, and identifying the erosion pit point cloud; (4) According to the erosion pit point cloud identified in step (3), using K-mean clustering algorithm for clustering and calculating the maximum depth of each pier erosion pit; (5) According to the erosion pit element classification value of all point clouds obtained in step (3), a curved surface reconstruction method based on erosion identification is proposed to perform high-precision curved surface reconstruction on the erosion pit.

2. The method according to claim 1, wherein the method is characterized by: The step (2) specifically comprises the following steps: (2.1) After posture correction processing and noise interference preprocessing of each measurement data, the measurement lines are merged; (2.2) The obtained water depth and position are processed, and the noise interference in the overlapping coverage range of two adjacent measurement lines is removed one by one; (2.3) Finally, the point cloud after denoising and merging is subjected to preliminary curved surface reconstruction by using the spherical rotation reconstruction method to form a triangular mesh.

3. The method according to claim 2, wherein the method is characterized by: In step (2.3), first, the density upper limit of the sonar point cloud is calculated for searching, then three points that do not contain other points are formed by specifying a radius p ball contact to form a seed triangle, and the ball is rotated around the edge until it touches the third point to form a new triangle, and finally all reachable edges are traversed to complete the preliminary reconstruction of the point cloud to form a triangular mesh.

4. The method of claim 1, wherein the method is characterized by: The step (3) specifically comprises the following steps: (3.1) Calculate the distance and semi-variance between all point clouds after preliminary reconstruction, select a Gaussian model as the theoretical variogram function to fit the semi-variance and distance function; (3.2) Set the search radius L and the skip radius R, and based on the Kriging interpolation, find the interpolation point cloud on the eight main directions within the radius range and outside the skip radius to obtain the interpolation point cloud on the eight elevation profiles; (3.3) the interpolated point cloud on the eight elevation profiles obtained according to step (3.2), on the principal direction profile of each point cloud according to the maximum elevation angle on the elevation profile and minimum elevation angle Computing an open angle and lower opening angle (3.4) Set the flatness t varying with the height of the point cloud, and calculate the elevation identifier of each profile according to the upper and lower opening angles obtained in step (3.3) (3.5) According to the principle of local ternary pattern, calculate the (+1) signs in the elevation identifier symbols on each point cloud 8 elevation profiles (-1) sign in the symbol Establishing scour identification search coordinates (3.6) the flushing identification search coordinates obtained according to step (3.5) Calculate the erosion pit element classification value θ of each point cloud; (3.7) Calculate the average value θ of the scour pit element classification values θ of all point clouds p as a threshold value for segmenting the scour pit point cloud; (3.8) Change the parameters of search radius, skip radius and flatness one by one in the value range by step, and perform identification calculation to identify the accuracy, when the identification accuracy meets the extraction requirements, perform erosion identification on the bridge underwater riverbed point cloud data according to the optimized parameters, and identify the erosion pit point cloud.

5. The method of claim 4, wherein the method is characterized by: The step (3.1) calculates the distance between all sample points and sorts them in ascending order, and then adopts a Gaussian model to functionally fit the relationship between the semi-variance value and the distance, and the general formula of the Gaussian model is: Where γ represents the semi-variance, h represents the distance, C0 is the block constant, C is the arch height, and a is the variation range, which are parameters that need to be preset when using Kriging interpolation.

6. The method of claim 5, wherein the method is characterized by: The flatness t varying with height in the step (3.4) is set as a single straight line model: where t is the flatness, t min is the minimum flatness, typically set to 0, t max is the maximum flatness, to be determined by tuning tests, z max is the highest elevation of the point cloud to be identified, z min is the lowest elevation of the point cloud to be identified.

7. The method of claim 6, wherein the method is characterized by: The calculated value of the scour pit element classification value θ defined in step (3.6) is as follows: first, calculate the scour identification search coordinates denotes the identification symbol of the eight profiles in the local ternary pattern, is the number of (+1) symbols, is the number of (-1) symbols; then calculate the mapping point of the distance from the mapping point (8,0) of the pit element, to define the scour pit element classification value θ of each point cloud, which is calculated as follows:

8. The method according to claim 7, wherein the method is characterized by: The spatial geometric features of the sonar point cloud obtained by statistical scanning in step (3.8) include: minimum average spacing and height variation coefficient; using a pre-test point cloud with the same spatial geometric features as the sonar point cloud randomly generated by a plurality of Gaussian amplitudes, the points below the horizontal plane are marked as scour pit point clouds, identifying the pre-test point cloud, and if the identification result meets the recall rate R e greater than 0.85, the F1 value greater than 0.89, the precision P r greater than 0.98, and the accuracy A c reach 0.93, then according to this tuning parameter, the scanned sonar point cloud is identified for scour, and the evaluation indexes are accuracy… c , recall rate R e , precision P r and F1 value are calculated as follows: Wherein TP is true positive, which means that the target point cloud is the sample point of the scour pit point cloud marked as the scour pit point cloud; FP is false positive, which means that the target point cloud is the sample point of the non-scour pit point cloud identified as the scour pit point cloud; TN is true negative, which means that the target point cloud is the sample point of the non-scour pit point cloud marked as the non-scour pit point cloud; FN is false negative, which means that the target point cloud is the sample point of the scour pit point cloud marked as the non-scour pit point cloud.

9. The method of claim 1, wherein the method is characterized by: The step (4) specifically comprises the following steps: (4.1) According to the scour pit point cloud identified in step (3), the mean and variance of the coordinates of the scour pit point cloud in the transverse direction are calculated, and denoising is performed; (4.2) Set the number of piers k in the range of the sonar point cloud, initialize k samples as the center a of the initial cluster k ; (4.3) For each point cloud p in the sonar point cloud, the distance of the point cloud p to the k cluster centers is calculated and attributed to the class corresponding to the cluster center with the smallest distance; The square Euclidean distance is used to calculate the distance of the point cloud p to the k cluster centers and attribute it to the class corresponding to the cluster center with the smallest distance, and the calculation formula of the square Euclidean distance is as follows: d(x, c) = (x - c) 2 Wherein d represents the distance, x represents the spatial coordinate row vector of the point cloud, and c represents the spatial coordinate row vector of the centroid of the cluster; (4.5) Repeat step (4.2) and step (4.3) until the iteration number reaches the upper limit, and obtain the clustering result of each pier scour pit point cloud; (4.6) For each cluster of pier scour pit point cloud, the minimum elevation is calculated to obtain the maximum depth of each scour pit of the sonar scanned bridge.

10. The method of claim 1, wherein the method is characterized by: The step (5) specifically comprises the following steps: (5.1) According to the triangular mesh obtained by the preliminary spherical rotation algorithm in step (2), the incenter of each triangle in the mesh is calculated; (5.2) According to the KNN algorithm, the neighboring points of the point cloud are obtained to perform covariance analysis to estimate the normal vector of each point cloud; Construct KD-Tree search structure of the scanning point cloud data, search the point cloud in the local neighborhood of each point cloud, calculate the centroid p of the field c Then calculate the covariance matrix of the point cloud p i as follows: M =∑(p i -p c )(p i -p c ) T In the formula, M represents the covariance matrix, and the eigenvalue of the covariance matrix is obtained according to the singular value decomposition method, wherein the smallest eigenvalue represents the normal vector of the point; (5.3) Calculate the tangent plane of each point in the point cloud; (5.4) Calculate the projection point of the incenter of the incircle of each adjacent triangle in the point cloud to the tangent plane of the point, and form an adjustable polygon by the projection point. The size of the adjustable polygon is adjusted by moving the position of the straight line on the point and the projection point, and the weight of the adjustment is the scour pit element classification value θ calculated in step (3); (5.5) Update the adjustable triangle position, move the vertex of the original triangular mesh to the position of the projection point, and sequentially traverse all the triangles to construct a new adjustable triangle; (5.6) According to the principle of the shortest diagonal, the space quadrilateral between the adjustable polygon and the adjustable triangle is segmented to obtain the triangulated patch; (5.7) Integrate the adjustable triangle, the adjustable polygon and the triangulated patch to obtain the finally reconstructed scour pit surface.

Images