Bridge tower swivel construction spatial form identification method based on three-dimensional laser point cloud

Abstract

The invention discloses a bridge tower rotation construction spatial form identification method based on three-dimensional laser point cloud, and the method comprises the steps: erecting a ground type three-dimensional laser scanner at a bridge tower rotation construction site, and collecting the point cloud data of a bridge tower rotation process; coordinate matrix transformation is carried out on the bridge tower rotation point cloud data acquired through multiple times of scanning, and coordinate system unification and correction of the multi-station bridge tower rotation point cloud data are realized; eliminating ground point clouds by adopting a parameter-optimized cloth simulation filtering algorithm, and reserving ground point clouds containing bridge towers; segmenting and extracting a bridge tower point cloud by adopting a clustering method according to the identified ground point cloud; because the segmented and extracted bridge tower point cloud contains noise, noise reduction is carried out by using a noise reduction method based on the adjacent point space relationship; the denoised bridge tower point cloud data volume is huge, and the analysis calculation amount is reduced by adopting uniform downsampling according to the analysis requirement; based on the down-sampled point cloud data of the bridge tower, a bridge tower point cloud boundary recognition algorithm is provided, and high-precision extraction of three-dimensional space coordinates of the bridge tower is completed.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent bridge construction, and in particular relates to a method for recognizing the spatial morphology of bridge tower rotation construction based on three-dimensional laser point clouds. Background Technology

[0002] Bridge rotation construction is a special method of bridge erection. Based on the rotation method, it can be divided into horizontal rotation and vertical rotation, and is commonly used in the construction of bridges spanning important passages such as railways, highways, or rivers. This method involves setting up a turntable or hinge point to assemble and construct the bridge structure in a position perpendicular to the direction of navigation or traffic. After the bridge structure is completed, a mechanical hydraulic device rotates the constructed bridge structure to the designed position to complete the construction. Bridge rotation construction can complete the construction process quickly without interrupting existing traffic, and is characterized by its high efficiency. It is widely used in the construction of bridge towers and main beams, and is an important bridge construction method.

[0003] The bridge tower rotation construction involves vertical rotation, resulting in a large rotational mass and high construction risks. The tower's alignment constantly changes during construction, necessitating a monitoring system to collect real-time spatial attitude information and dynamically adjust rotation parameters to ensure the tower's alignment meets design goals and guarantees precise positioning. Traditional monitoring methods involve placing reflective targets on the tower top and using a total station to measure the rotation angle. However, these methods rely on manual measurement, have low automation, and only collect rotation angle data at single points, providing limited monitoring information and failing to reflect changes in the tower's spatial attitude during the rotation process, thus increasing the overall risk.

[0004] 3D laser scanning technology is a technique based on laser ranging principles to acquire high-precision 3D spatial information. It can quickly and non-contactly measure objects, collect massive amounts of high-density 3D point cloud data, and comprehensively and accurately record the 3D shape and surface features of objects, effectively replacing traditional surveying equipment such as total stations. While 3D laser scanning technology can automatically acquire point cloud data at bridge tower rotation construction sites, achieving spatial morphology recognition of bridge towers faces the following challenges: First, the coordinate systems of point cloud data collected from multiple laser scanning stations are not unified, and there is an angle between the bridge axis and the coordinate axes, making direct analysis of point cloud data collected from multiple stations impossible. Second, the construction site environment for bridge tower rotation is complex, and the collected point clouds contain bridge decks, construction machinery, and other debris, resulting in redundant point cloud data, requiring solutions to bridge tower point cloud segmentation. Finally, there is a lack of research on spatial morphology extraction from bridge tower point clouds, which makes spatial morphology recognition during bridge tower rotation difficult. Therefore, applying 3D laser scanning technology to spatial morphology recognition during bridge tower rotation construction requires addressing these issues. Summary of the Invention

[0005] Purpose of the invention: The purpose of this invention is to provide a method for recognizing the spatial morphology of bridge tower rotation construction based on three-dimensional laser point clouds. This method can automatically collect point clouds at the bridge tower rotation construction site, realize the segmentation of the bridge tower point cloud and the extraction of its three-dimensional spatial morphology. It has strong applicability, can solve the shortcomings of traditional measurement methods, and can provide data support for the construction control of bridge tower rotation.

[0006] Technical Solution: To achieve the above objectives, this invention discloses a method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds. The method includes the following steps:

[0007] (1) During the bridge tower rotation construction, a ground-based three-dimensional laser scanner was set up to collect the original point cloud of the bridge tower rotation every 30 minutes to realize the acquisition of point cloud data throughout the entire bridge tower rotation process.

[0008] (2) Perform coordinate matrix transformation on the original point cloud data of the bridge tower rotation construction site collected from multiple stations to realize coordinate system unification and correction of point cloud data collected from multiple scanning stations;

[0009] (3) Based on the original point cloud data of the bridge tower rotation construction site preliminarily processed in step (2), the parameter-optimized cloth filtering algorithm is used to identify all non-ground points containing the bridge tower point cloud and filter out all ground point clouds.

[0010] (4) Based on all non-ground points containing bridge tower point clouds extracted in step (3), the DBSCAN point cloud density clustering method is used to identify and extract bridge tower point clouds.

[0011] (5) Based on the bridge tower point cloud segmented and extracted in step (4), noise reduction is performed using a noise reduction method based on the spatial relationship of neighboring points to remove the noise contained in the bridge tower point cloud.

[0012] (6) Based on the noise-reduced bridge tower point cloud data obtained in step (5), the noise-reduced bridge tower point cloud data is thinned out using a uniform downsampling method.

[0013] (7) Based on the downsampled bridge tower point cloud data obtained in step (6), the bridge tower point cloud boundary recognition algorithm is used to extract the spatial morphology of the bridge tower.

[0014] Furthermore, step (2) specifically includes the following steps:

[0015] (2.1) Based on the conditions of the bridge tower rotation construction site, set up common targets and extract the center coordinates of the common targets of multiple scanning stations. Construct a coordinate matrix based on the center coordinates of the common targets and solve the coordinate transformation matrix between the coordinate matrices of multiple scanning stations. Apply the coordinate transformation matrix to unify the coordinate system of the point cloud collected by multiple scanning stations.

[0016] (2.2) Using the symmetrical cables between the bridge towers as a reference, the RANSAC algorithm is used to fit the axis equation of the bridge, calculate the rotation angle of the original point cloud and rotate the original point cloud to realize the correction of the axis of the original point cloud coordinate system.

[0017] Furthermore, step (3) specifically includes the following steps:

[0018] (3.1) The original point cloud collected at the bridge tower rotation construction site was flipped;

[0019] (3.2) Initialize the cloth grid, optimize the grid resolution parameters and determine the number of grid nodes;

[0020] (3.3) Project the laser points and grid points onto the same horizontal plane, match the grid points with the laser points one by one, and record the elevation of the laser points;

[0021] (3.4) When grid points move due to gravity, compare the elevation of the grid point with the elevation of the corresponding laser point. If the elevation is less than or equal to the elevation of the laser point, replace the corresponding laser point with the grid point and calculate the position of all grid points affected by the movement of neighboring nodes.

[0022] (3.5) Repeat step (3.4) and terminate the simulation process when the maximum elevation change of all grid points is less than the set threshold or exceeds the set maximum number of iterations.

[0023] (3.6) Calculate the distance between grid points and corresponding laser points, and divide the point cloud into ground point cloud and non-ground point cloud according to the set distance threshold.

[0024] Furthermore, step (4) specifically includes the following steps:

[0025] (4.1) Scan all non-ground point clouds. If the number of points within a set radius R of a sample point is greater than or equal to the set minimum number of points, list it as a core point and form a temporary cluster of points within a set radius R of the core point.

[0026] (4.2) For each temporary cluster, check whether the points in it are core points. If they are, merge the temporary cluster corresponding to the point with the current temporary cluster to obtain a new temporary cluster. Repeat this operation until every point in the current temporary cluster is a core point, and upgrade the temporary cluster to a cluster. Continue to perform the same merging operation on the remaining temporary clusters until all temporary clusters are processed, thereby effectively segmenting the bridge tower point cloud from the non-ground point cloud.

[0027] Furthermore, step (5) specifically includes the following steps:

[0028] (5.1) The segmented and extracted bridge tower point cloud is rasterized and divided into blocks. The resolution of the raster is determined and the point cloud is divided into units with length, width and height of Δx×Δy×Δz. The three-dimensional coordinates of the bridge tower point cloud are converted into raster coordinate index g. x g y g z :

[0029]

[0030] Where, x i y i z i Let x be the three-dimensional coordinates of the point cloud of the bridge tower. min y min z min The minimum three-dimensional coordinates of the point cloud for the bridge tower;

[0031] (5.2) Retrieve each point in the grid and perform spatial plane detection according to the kd-tree algorithm of the point cloud, and filter out the grids that fail the spatial plane detection;

[0032] (5.3) Within the grid that has not passed the spatial plane detection, select each point P in sequence. i Search point P based on Euclidean distance i The 8 nearest neighbors P iq As P i The set of decision points, q = 1, 2, ... 8;

[0033] (5.4) According to P iq Determine the plane N by its spatial coordinates, and determine point P. i If the vertical distance to plane N exceeds the set noise threshold ε, then the noise is deleted. Follow the steps (5.3)-(5.4) to search for other points in the grid that have not passed the spatial plane detection until all points in the grid have been searched.

[0034] Furthermore, step (6) specifically includes the following steps:

[0035] (6.1) Set the retrieval radius r, and randomly select an initial point from the bridge tower point cloud to add to the output point set P. j Calculate the remaining points P i With output point set P j The distance d between all points:

[0036]

[0037] Where, x i y i z i It is point P i The coordinates, xj y j z j It is point P j The coordinates;

[0038] (6.2) Determine the relationship between distance d and search radius r. When distance d is greater than or equal to search radius r, move point P. i Add to output point set P j Otherwise, point P will be used. i Delete, ensuring point set P j The distance between all points is greater than or equal to the retrieval radius r, thus achieving uniform downsampling of the point cloud.

[0039] Furthermore, step (7) specifically includes the following steps:

[0040] (7.1) Project the downsampled bridge tower point cloud onto the YOZ plane and establish the mapping relationship between the three-dimensional bridge tower point cloud and the two-dimensional bridge tower point cloud. Set the two-dimensional bridge tower point cloud after projection onto the YOZ plane as the point set S and set the parameter α.

[0041] (7.2) Construct Delaunay partition based on point set S, establish Delaunay triangulation. If the length of a side of a triangle is greater than 2α, delete the triangle. For each side of the remaining triangle, if a circle with radius α passing through two points of a side of a triangle contains another point of the triangle, delete the triangle as well.

[0042] (7.3) Solve the edge of the remaining Delaunay triangulation to form α-shape, that is, the boundary B2 of the two-dimensional bridge tower point cloud. Return the boundary B2 of the two-dimensional bridge tower point cloud to the three-dimensional space according to the mapping relationship to obtain the boundary B3 of the three-dimensional bridge tower point cloud.

[0043] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0044] This invention utilizes a 3D laser scanner to automatically acquire high-density, high-quality point cloud data from the bridge tower rotation construction site. This effectively overcomes the limitations of traditional single-point measurements. Based on a common target deployed at the construction site, it unifies the coordinate system of multi-site point cloud data. The RANSAC algorithm is used to fit the bridge's axis equation to correct the point cloud coordinate system. Simultaneously, this invention achieves automated segmentation of the bridge tower point cloud data. A material filtering algorithm separates the ground point cloud from the non-ground point cloud, and the DBSCAN algorithm is further used to segment the bridge tower point cloud from the non-ground point cloud. Based on the segmented and extracted bridge tower point cloud, point cloud denoising and boundary recognition algorithms are developed to extract the 3D spatial morphology during the bridge tower rotation process, providing effective data support for construction control at the bridge tower rotation construction site. This invention enables rapid and high-precision extraction of the spatial morphology of bridge tower rotation. Attached Figure Description

[0045] Figure 1 A flowchart illustrating the implementation of this invention;

[0046] Figure 2 To unify and correct the cloud coordinates of the bridge tower rotation construction points;

[0047] Figure 3 Classification of point clouds during bridge tower rotation construction;

[0048] Figure 4 Segmenting the bridge tower point cloud using the DBSCAN algorithm;

[0049] Figure 5 To reduce noise in cloud points for bridge towers;

[0050] Figure 6 Extracting the three-dimensional spatial morphology of bridge tower point clouds. Detailed Implementation

[0051] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0052] This invention primarily serves the technical field of bridge tower spatial morphology recognition based on three-dimensional laser scanning, and the applied object is three-dimensional point cloud data collected at the bridge tower rotation construction site.

[0053] like Figure 1 As shown, a method for recognizing the spatial morphology of bridge tower rotation construction based on three-dimensional laser point clouds includes the following steps:

[0054] (1) During the bridge tower rotation construction, a ground-based three-dimensional laser scanner was set up to collect the original point cloud of the bridge tower rotation every 30 minutes to realize the acquisition of point cloud data throughout the entire bridge tower rotation process.

[0055] (2) Perform coordinate matrix transformation on the original point cloud data of the bridge tower rotation construction site collected from multiple stations to realize coordinate system unification and correction of point cloud data collected from multiple scanning stations;

[0056] like Figure 2 As shown, step (2) specifically includes the following steps:

[0057] (2.1) Based on the conditions of the bridge tower rotation construction site, set up common targets and extract the center coordinates of the common targets of multiple scanning stations. Construct a coordinate matrix based on the center coordinates of the common targets and solve the coordinate transformation matrix between the coordinate matrices of multiple scanning stations. Apply the coordinate transformation matrix to realize the coordinate system of the point cloud collected by multiple scanning stations.

[0058] (2.2) Using the symmetrical cables between the bridge towers as a reference, the RANSAC algorithm is used to fit the axis equation of the bridge, calculate the rotation angle of the original point cloud and rotate the original point cloud to realize the correction of the axis of the original point cloud coordinate system.

[0059] (3) Based on the original point cloud data of the bridge tower rotation construction site preliminarily processed in step (2), the parameter-optimized cloth filtering algorithm is used to identify all non-ground points containing the bridge tower point cloud and filter out all ground point clouds.

[0060] like Figure 3 As shown, step (3) specifically includes the following steps:

[0061] (3.1) The original point cloud collected at the bridge tower rotation construction site was flipped;

[0062] (3.2) Initialize the cloth grid, optimize the grid resolution parameters and determine the number of grid nodes;

[0063] (3.3) Project the laser point and the grid point onto the same horizontal plane, match the grid point with the laser point one by one, and record the elevation of the laser point.

[0064] (3.4) When grid points move due to gravity, compare the elevation of the grid point with the elevation of the corresponding laser point. If the elevation is less than or equal to the elevation of the laser point, replace the corresponding laser point with the grid point and calculate the position of all grid points affected by the movement of neighboring nodes.

[0065] (3.5) Repeat step (3.4) and terminate the simulation process when the maximum elevation change of all grid points is less than the set threshold or exceeds the set maximum number of iterations.

[0066] (3.6) Calculate the distance between grid points and corresponding laser points, and divide the point cloud into ground point cloud and non-ground point cloud according to the set distance threshold.

[0067] (4) Based on all non-ground points containing bridge tower point clouds extracted in step (3), the DBSCAN point cloud density clustering method is used to identify and extract bridge tower point clouds.

[0068] like Figure 4 As shown, step (4) specifically includes the following steps:

[0069] (4.1) Scan all non-ground point clouds. If the number of points within a set radius R of a sample point is greater than or equal to the set minimum number of points, list it as a core point and form a temporary cluster of points within a set radius R of the core point.

[0070] (4.2) For each temporary cluster, check whether the points in it are core points. If they are, merge the temporary cluster corresponding to the point with the current temporary cluster to obtain a new temporary cluster. Repeat this operation until every point in the current temporary cluster is a core point, and upgrade the temporary cluster to a cluster. Continue to perform the same merging operation on the remaining temporary clusters until all temporary clusters are processed, thereby effectively segmenting the bridge tower point cloud from the non-ground point cloud.

[0071] (5) Based on the bridge tower point cloud segmented and extracted in step (4), noise reduction is performed using a noise reduction method based on the spatial relationship of neighboring points to remove the noise contained in the bridge tower point cloud.

[0072] like Figure 5 As shown, step (5) specifically includes the following steps:

[0073] (5.1) The segmented and extracted bridge tower point cloud is rasterized and divided into blocks. The resolution of the raster is determined and the point cloud is divided into units with length, width and height of Δx×Δy×Δz. The three-dimensional coordinates of the bridge tower point cloud are converted into raster coordinate index g. x g y g z :

[0074]

[0075] Where, x i y i z i Let x be the three-dimensional coordinates of the point cloud of the bridge tower. min y min z min The minimum three-dimensional coordinates of the point cloud for the bridge tower.

[0076] (5.2) Retrieve each point in the grid and perform spatial plane detection according to the kd-tree algorithm of the point cloud, and filter out the grids that fail the spatial plane detection;

[0077] (5.3) Within the grid that has not passed the spatial plane detection, select each point P in sequence. i Search point P based on Euclidean distance i The 8 nearest neighbors P iq q = 1, 2, ... 8, as P i The set of decision points;

[0078] (5.4) According to P iq Determine the plane N by its spatial coordinates, and determine point P. i If the vertical distance to plane N exceeds the set noise threshold ε, then the noise is deleted. Follow the steps (5.3)-(5.4) to search for other points in the grid that have not passed the spatial plane detection until all points in the grid have been searched.

[0079] (6) Based on the noise-reduced bridge tower point cloud data obtained in step (5), the noise-reduced bridge tower point cloud data is thinned using a uniform downsampling method to reduce the amount of analysis and calculation of the bridge tower point cloud.

[0080] (6.1) Set the retrieval radius r, and randomly select an initial point from the bridge tower point cloud to add to the output point set P. j Calculate the remaining points P i With output point set P j The distance d between all points:

[0081]

[0082] Where, x i y i z i It is point P i The coordinates, x j y j z j It is point P j The coordinates;

[0083] (6.2) Determine the relationship between distance d and search radius r. When distance d is greater than or equal to search radius r, move point P. i Add to output point set P j Otherwise, point P will be used. i Delete, ensuring point set P j The distance between all points is greater than or equal to the retrieval radius r, thus achieving uniform downsampling of the point cloud.

[0084] (7) Based on the downsampled bridge tower point cloud data obtained in step (6), develop a bridge tower point cloud boundary recognition algorithm to complete the high-precision extraction of bridge tower spatial morphology.

[0085] like Figure 6 As shown, step (7) specifically includes the following steps:

[0086] (7.1) Project the downsampled bridge tower point cloud onto the YOZ plane and establish the mapping relationship between the three-dimensional bridge tower point cloud and the two-dimensional bridge tower point cloud. Set the two-dimensional bridge tower point cloud after projection onto the YOZ plane as the point set S and set the parameter α.

[0087] (7.2) Construct Delaunay partition based on point set S, establish Delaunay triangulation. If the length of a side of a triangle is greater than 2α, delete the triangle. For each side of the remaining triangle, if a circle with radius α passing through two points of a side of a triangle contains another point of the triangle, delete the triangle as well.

[0088] (7.3) Solve for the edges of the remaining Delaunay triangulation to form the α-shape, i.e. the boundary B2 of the two-dimensional bridge tower point cloud. Return the boundary B2 of the two-dimensional bridge tower point cloud to the three-dimensional space according to the mapping relationship to obtain the boundary B3 of the three-dimensional bridge tower point cloud.

Claims

1. A method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds, characterized in that, The method includes the following steps: (1) During the bridge tower rotation construction, a ground-based three-dimensional laser scanner was set up to collect the original point cloud of the bridge tower rotation every 30 minutes to realize the acquisition of point cloud data throughout the entire bridge tower rotation process. (2) Perform coordinate matrix transformation on the original point cloud data of the bridge tower rotation construction site collected from multiple stations to realize coordinate system unification and correction of point cloud data collected from multiple scanning stations; (3) Based on the original point cloud data of the bridge tower rotation construction site preliminarily processed in step (2), the parameter-optimized cloth filtering algorithm is used to identify all non-ground points containing the bridge tower point cloud and filter out all ground point clouds. (4) Based on all non-ground points containing bridge tower point clouds extracted in step (3), the DBSCAN point cloud density clustering method is used to identify and extract bridge tower point clouds. (5) Based on the bridge tower point cloud segmented and extracted in step (4), noise reduction is performed using a noise reduction method based on the spatial relationship of neighboring points to remove the noise contained in the bridge tower point cloud. (6) Based on the noise-reduced bridge tower point cloud data obtained in step (5), the noise-reduced bridge tower point cloud data is thinned out using a uniform downsampling method. (7) Based on the downsampled bridge tower point cloud data obtained in step (6), the bridge tower point cloud boundary recognition algorithm is used to extract the spatial morphology of the bridge tower.

2. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (2) specifically includes the following steps: (2.1) Based on the conditions of the bridge tower rotation construction site, set up common targets and extract the center coordinates of the common targets of multiple scanning stations. Construct a coordinate matrix based on the center coordinates of the common targets and solve the coordinate transformation matrix between the coordinate matrices of multiple scanning stations. Apply the coordinate transformation matrix to unify the coordinate system of the point cloud collected by multiple scanning stations. (2.2) Using the symmetrical cables between the bridge towers as a reference, the RANSAC algorithm is used to fit the axis equation of the bridge, calculate the rotation angle of the original point cloud and rotate the original point cloud to realize the correction of the axis of the original point cloud coordinate system.

3. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (3) specifically includes the following steps: (3.1) The original point cloud collected at the bridge tower rotation construction site was flipped; (3.2) Initialize the cloth grid, optimize the grid resolution parameters and determine the number of grid nodes; (3.3) Project the laser points and grid points onto the same horizontal plane, match the grid points with the laser points one by one, and record the elevation of the laser points; (3.4) When grid points move due to gravity, compare the elevation of the grid point with the elevation of the corresponding laser point. If the elevation is less than or equal to the elevation of the laser point, replace the corresponding laser point with the grid point and calculate the position of all grid points affected by the movement of neighboring nodes. (3.5) Repeat step (3.4) and terminate the simulation process when the maximum elevation change of all grid points is less than the set threshold or exceeds the set maximum number of iterations. (3.6) Calculate the distance between grid points and corresponding laser points, and divide the point cloud into ground point cloud and non-ground point cloud according to the set distance threshold.

4. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (4) specifically includes the following steps: (4.1) Scan all non-ground point clouds. If the number of points within a set radius R of a sample point is greater than or equal to the set minimum number of points, list it as a core point and form a temporary cluster of points within a set radius R of the core point. (4.2) For each temporary cluster, check whether the points in it are core points. If they are, merge the temporary cluster corresponding to the point with the current temporary cluster to obtain a new temporary cluster. Repeat this operation until every point in the current temporary cluster is a core point, and upgrade the temporary cluster to a cluster. Continue to perform the same merging operation on the remaining temporary clusters until all temporary clusters are processed, thereby effectively segmenting the bridge tower point cloud from the non-ground point cloud.

5. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (5) specifically includes the following steps: (5.1) The segmented and extracted bridge tower point cloud is rasterized and divided into blocks. The resolution of the raster is determined and the point cloud is divided into units with length, width and height of Δx×Δy×Δz. The three-dimensional coordinates of the bridge tower point cloud are converted into raster coordinate index g. x g y g z : Where, x i y i z i Let x be the three-dimensional coordinates of the point cloud of the bridge tower. min y min z min The minimum three-dimensional coordinates of the point cloud for the bridge tower; (5.2) Retrieve each point in the grid and perform spatial plane detection according to the kd-tree algorithm of the point cloud, and filter out the grids that fail the spatial plane detection; (5.3) Within the grid that has not passed the spatial plane detection, select each point P in sequence. i Search point P based on Euclidean distance i The 8 nearest neighbors P iq As P i The set of decision points, q = 1, 2, ... 8; (5.4) According to P iq Determine the plane N by its spatial coordinates, and determine point P. i If the vertical distance to plane N exceeds the set noise threshold ε, then the noise is deleted. Follow the steps (5.3)-(5.4) to search for other points in the grid that have not passed the spatial plane detection until all points in the grid have been searched.

6. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (6) specifically includes the following steps: (6.1) Set the retrieval radius r, and randomly select an initial point from the bridge tower point cloud to add to the output point set P. j Calculate the remaining points P i With output point set P j The distance d between all points: Where, x i y i z i It is point P i The coordinates, x j y j z j It is point P j The coordinates; (6.2) Determine the relationship between distance d and search radius r. When distance d is greater than or equal to search radius r, move point P. i Add to output point set P j Otherwise, point P will be used. i Delete, ensuring point set P j The distance between all points is greater than or equal to the retrieval radius r, thus achieving uniform downsampling of the point cloud.

7. The method for spatial morphology recognition during bridge tower rotation construction based on three-dimensional laser point clouds according to claim 1, characterized in that: Step (7) specifically includes the following steps: (7.1) Project the downsampled bridge tower point cloud onto the YOZ plane and establish the mapping relationship between the three-dimensional bridge tower point cloud and the two-dimensional bridge tower point cloud. Set the two-dimensional bridge tower point cloud after projection onto the YOZ plane as the point set S and set the parameter α. (7.2) Construct Delaunay partition based on point set S, establish Delaunay triangulation. If the length of a side of a triangle is greater than 2α, delete the triangle. For each side of the remaining triangle, if a circle with radius α passing through two points of a side of a triangle contains another point of the triangle, delete the triangle as well. (7.3) Solve the edge of the remaining Delaunay triangulation to form α-shape, that is, the boundary B2 of the two-dimensional bridge tower point cloud. Return the boundary B2 of the two-dimensional bridge tower point cloud to the three-dimensional space according to the mapping relationship to obtain the boundary B3 of the three-dimensional bridge tower point cloud.

Images