Method for automatically generating geometric finite element model of ancient building structure based on three-dimensional point cloud

Abstract

The invention discloses an ancient building structure geometric finite element model automatic generation method based on a three-dimensional point cloud. The method comprises the steps that the three-dimensional point cloud of an ancient building structure is acquired; segmenting the three-dimensional point cloud to obtain an independent component point cloud; extracting geometric information of the component point cloud; and constructing the geometric finite element model of the ancient building structure based on the geometric information. The three-dimensional point cloud data can completely record overall and local details of the historic building, and comprehensive analysis and protection are facilitated.

Description

Technical Field

[0001] The present invention belongs to the technical field of the reconstruction of finite element models and structural health monitoring of ancient buildings, and particularly relates to a method for automatically generating a geometric finite element model of an ancient building structure based on three-dimensional point clouds. Background Art

[0002] Ancient buildings are an important part of the traditional culture of the Chinese nation. They are not only a witness to history but also a crystallization of art and science. These buildings have withstood the wind and rain for hundreds or even thousands of years, witnessing countless historical changes and social developments. However, during their long service life, under the influence of external loads such as climate change, human activities, and earthquakes, as well as internal factors such as wood aging and fatigue, ancient buildings will show varying degrees of damage and destruction, posing potential safety hazards, which has given rise to the engineering needs for structural health monitoring and safety assessment. Finite element models are indispensable tools in modern engineering analysis, especially playing an important role in aspects such as the structural health monitoring, seismic analysis, and safety assessment of ancient buildings. These models can simulate the responses of ancient buildings under the action of various external and internal factors, thereby predicting possible damage and destruction. However, in practical engineering applications, due to the complexity and uniqueness of ancient buildings and the possible loss of original design drawings, researchers are faced with the challenge of reconstructing accurate models. Traditional methods, such as relying on on-site measurements and manual drawing, are not only time-consuming and laborious but also prone to errors. The limitations of these methods have given rise to the need for more efficient and reliable finite element model generation methods. Three-dimensional laser scanning technology is an advanced measurement technology that can accurately reconstruct the three-dimensional data and models of targets through precise laser scanning. This technology can perform comprehensive or partial high-precision measurements on targets, capture multi-dimensional information such as lines, surfaces, volumes, and spaces, and generate a data set called "point cloud". Point cloud is a set of points in three-dimensional space, usually containing position information, and sometimes also including color information and object reflection surface intensity information. Point cloud data can comprehensively and accurately record the shape, size, and structure of buildings, significantly improving the speed and accuracy of model generation, and providing strong technical support for the protection and restoration of ancient buildings. However, since the original data collected by laser scanners is unstructured and contains a large amount of complex information, it has become particularly important to develop a fast and automated method to interpret the structural information in these point clouds. Summary of the Invention

[0003] To solve the above technical problems, the present invention proposes a method for automatically generating a geometric finite element model of an ancient building structure based on three-dimensional point clouds, which solves the problems of the loss of ancient building drawings and the difficulty of model reconstruction. The established finite element model can be used for subsequent structural analysis and health maintenance.

[0004] The present invention provides a method for automatically generating a geometric finite element model of an ancient building structure based on 3D point clouds, including:

[0005] Obtaining the 3D point cloud of the ancient building structure;

[0006] Segmenting the 3D point cloud to obtain independent component point clouds;

[0007] Extracting the geometric information of the component point clouds;

[0008] Constructing the geometric finite element model of the ancient building structure based on the geometric information.

[0009] Optionally, segmenting the 3D point cloud to obtain independent component point clouds includes:

[0010] Using the region growing method to segment the surface point cloud of the component;

[0011] Classifying the surface point cloud of the component to obtain independent component point clouds, where the component point clouds include: cylindrical component point clouds and hexahedral component point clouds.

[0012] Optionally, using the region growing method to segment the surface point cloud of the component includes:

[0013] S1. Setting the number of points in the minimum point cluster, the number of points in the maximum point cluster, the smoothing threshold, and the curvature threshold;

[0014] S2. Obtaining the curvature values of the points in the surface point cloud of the component, sorting the curvature values, and selecting the point with the minimum curvature value as the current seed point;

[0015] S3. Adding the current seed point to the seed point sequence and searching for the neighborhood points of the current seed point;

[0016] S4. Comparing the angle between the normal of the neighborhood point and the normal of the current seed point, determining whether the angle satisfies the smoothing threshold, and adding the current seed point that satisfies the smoothing threshold to the clustering region;

[0017] S5. Calculating the curvature values of the neighborhood points, adding the neighborhood points that satisfy the curvature threshold to the seed sequence, and updating the current seed point;

[0018] S6. Repeating S3 - S5 until the seed sequence is empty or the number of points in the point cluster generated by the remaining point cloud does not satisfy the number of points in the minimum point cluster.

[0019] Optionally, the method for selecting the point with the minimum curvature value as the current seed point is:

[0020] Given a set of point clouds P = {P1, P2, …, P n};

[0021] Calculating the point Pi The mean vector of the k neighborhood points;

[0022] According to the mean vector, calculate the covariance matrix formed by the k neighborhood points of point P i ;

[0023] Calculate the eigenvalues and corresponding eigenvectors of the covariance matrix;

[0024] According to the eigenvalues and corresponding eigenvectors, obtain the curvature value.

[0025] Optionally, extracting the geometric information of the cylinder component point cloud includes:

[0026] Using the PCA principal component analysis method to calculate the point cloud covariance eigenvector to obtain the center line direction of the cylinder component;

[0027] Project the cylinder component point cloud onto the cross-section XOY, and use the least squares method for point cloud fitting to obtain the center line position and radius of the cylinder component;

[0028] Based on the center line position and radius of the cylinder component, use projection and grid division method to obtain the three-dimensional plane equations of the upper and lower end faces of the cylinder component.

[0029] Optionally, using the projection and grid division method to obtain the three-dimensional plane equations of the upper and lower end faces of the cylinder component includes:

[0030] S1. Project the cylinder component point cloud onto the longitudinal section XOZ, and obtain the boundary point cloud through the grid division method;

[0031] S2. Based on the boundary point cloud, use the RANSAC linear iterative fitting segmentation to obtain the two-dimensional linear equation of the projection of the cylinder end face on the longitudinal section XOZ;

[0032] S3. According to the end face projection two-dimensional linear direction vector and the projection plane normal vector, obtain the end face normal vector, and then combine the coordinates of any point on the plane to obtain the three-dimensional plane equation of the end face.

[0033] Optionally, projecting the cylinder component point cloud onto the longitudinal section XOZ and obtaining the boundary point cloud through the grid division method includes:

[0034] Project the cylinder component point cloud onto the longitudinal section XOZ and perform grid division to obtain the grid type, where the grid type includes: solid hole grid and empty hole grid;

[0035] Detect the grids adjacent to the current solid hole grid. If there is at least one empty hole grid, define the current solid hole grid as the boundary grid;

[0036] According to the boundary grid, obtain the boundary point cloud.

[0037] Optionally, based on the boundary point cloud, the two-dimensional linear equation of the projection of the cylindrical end face point cloud on the longitudinal section XOZ obtained by RANSAC linear iterative fitting segmentation includes:

[0038] S1. Randomly select two points from the boundary point cloud to obtain a linear equation;

[0039] S2. Calculate the distances from the remaining points in the boundary point cloud to the linear equation, and count the number of inliers according to the distances;

[0040] S3. Based on the inliers, obtain the model parameters;

[0041] S4. Remove the inliers from the boundary point cloud to obtain a new point cloud;

[0042] S5. Repeat S1 - S4 until all the linear equations, i.e., the two-dimensional linear equation, are found.

[0043] Optionally, extracting the geometric information of the hexahedron component point cloud includes:

[0044] Using the RANSAC plane fitting method to obtain the main plane equation of the hexahedron component point cloud;

[0045] Using the projection and grid division method to obtain the side plane equations of the hexahedron component;

[0046] Based on the main plane equation and the side plane equations, obtain the geometric information of the hexahedron component point cloud;

[0047] Among them, using the RANSAC plane fitting method to obtain the main plane equation of the hexahedron component point cloud includes:

[0048] S1. Randomly select three points from the main plane point cloud to obtain a plane equation;

[0049] S2. Calculate the distances from the remaining points in the boundary point cloud to the plane equation, and count the number of inliers according to the distances;

[0050] S3. Based on the inliers, obtain the model parameters;

[0051] S4. Remove the inliers from the boundary point cloud to obtain a new point cloud;

[0052] S5. Repeat S1 - S4 until all the plane equations, i.e., the main plane equation, are found.

[0053] Using the projection and grid division method to obtain the side plane equations of the hexahedron component includes:

[0054] Project the point cloud of the hexahedron component onto the main plane, and obtain the boundary point cloud through the grid division method;

[0055] Based on the boundary point cloud, use the RANSAC linear iterative fitting segmentation to obtain the two-dimensional linear equation of the projection of the hexahedron side on the main plane;

[0056] Obtain the side normal vector according to the two-dimensional linear direction vector of the end face projection and the normal vector of the projection plane, and then combine the coordinates of any point on the plane to obtain the three-dimensional plane equation of the side.

[0057] Optionally, automatically output the geometric finite element model, and realize the parametric modeling, analysis, and visualization of the finite element model based on the ANSYS APDL language.

[0058] Compared with the prior art, the present invention has the following advantages and technical effects:

[0059] 1. The method proposed by the present invention combines the three-dimensional point cloud processing technology and the ANSYS APDL parametric modeling technology, and can automatically and quickly construct a three-dimensional geometric finite element model of the ancient building structure from the point cloud data collected by the three-dimensional laser scanner, reducing the time and labor costs of manual modeling and improving the work efficiency;

[0060] 2. The three-dimensional point cloud data of the present invention can completely record the overall and local details of the ancient building, which is helpful for comprehensive analysis and protection;

[0061] 3. The finite element model generated by the present invention can be used for structural mechanics analysis, evaluate the safety and stability of ancient buildings, formulate more scientific repair and protection plans, and extend the lifespan of ancient buildings. Description of the Drawings

[0062] The drawings constituting a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0063] Figure 1 is a flowchart of a method for automatically generating a geometric finite element model of an ancient building structure based on three-dimensional point cloud according to an embodiment of the present invention;

[0064] Figure 2 is a schematic diagram of component segmentation according to an embodiment of the present invention;

[0065] Figure 3 is a schematic diagram of the description of geometric information of various components according to an embodiment of the present invention;

[0066] Figure 4 is a schematic diagram of the method for extracting geometric information of various components according to an embodiment of the present invention;

[0067] Figure 5 It is a schematic diagram of a case flow of automatically generating a geometric finite element model of a vertical brick-wood structure based on a three-dimensional point cloud according to an embodiment of the present invention. DETAILED DESCRIPTION

[0068] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0069] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0070] This embodiment provides a method for automatically generating a geometric finite element model of an ancient building structure based on a three-dimensional point cloud. Figure 1 As shown, the specific steps include:

[0071] (1) Obtaining a point cloud model of the structure: Using a 3D laser scanner to scan the structure in all directions to obtain a 3D point cloud model of the structure;

[0072] (2) Component segmentation: The surface of the component is separated using a regional segmentation algorithm, and then the point cloud is grouped according to spatial position and size characteristics to obtain independent component point clouds;

[0073] (3) Geometric information extraction: Based on the component point cloud obtained in step (2), the component geometric information is obtained for modeling through PCA principal component analysis, least squares method, grid division method, RANSAC point cloud fitting, etc.;

[0074] (4) Automatically output the geometric finite element model: parametric modeling, analysis, and visualization of the finite element model are realized based on the ANSYS APDL language. Through the geometric and position information of point cloud processing, analysis, and recognition in step (3), the APDL command file for component parametric modeling is automatically generated to obtain the geometric finite element model of the ancient building structure based on the three-dimensional point cloud.

[0075] Furthermore, in step (1), the point cloud is registered to the global coordinate system using the software provided by the 3D laser scanner, and outliers in the point cloud are filtered and down-sampled.

[0076] Furthermore, in step (2), segmenting the three-dimensional point cloud to obtain independent component point clouds includes:

[0077] The point cloud of the component surface is segmented using the region growing method;

[0078] Classify the point cloud on the surface of the component to obtain independent component point clouds, where the component point clouds include: cylindrical component point clouds and hexahedral component point clouds.

[0079] Specifically, first use the region growing method to segment the point cloud on the surface of the component, determine the shape type of the surface point cloud through the relative relationship of covariance eigenvalues, determine the component type to which the point cloud belongs through the relative relationship of dimensions, and finally group the surface point cloud and determine the contact relationship through the point cloud range and relative position relationship to obtain independent component point clouds, as Figure 2 shown.

[0080] Furthermore, using the region growing method to segment the point cloud on the surface of the component includes:

[0081] S1. Set the number of points n of the minimum point cluster min , the number of points n of the maximum point cluster max , the smoothing threshold θ T , the curvature threshold κ T ;

[0082] S2. Obtain the curvature values of the points in the point cloud on the surface of the component, sort the curvature values, and select the point with the minimum curvature value as the current seed point;

[0083] S3. Add the current seed point to the seed point sequence Q and search for the neighborhood points of the current seed point;

[0084] S4. Compare the angle θ j between the normal of the neighborhood point and the normal of the current seed point, and judge whether the angle satisfies the smoothing threshold θ T , and add the current seed point that satisfies the smoothing threshold to the clustering region R;

[0085] S5. Calculate the curvature value κ j of the neighborhood point, and add the neighborhood point that satisfies the curvature threshold κ T to the seed sequence Q and update the current seed point;

[0086] S6. Repeat S3 - S5 until the seed sequence is empty or the number of points in the point cluster generated by the remaining point cloud does not meet the number of points n of the minimum point cluster min .

[0087] Specifically, the method of selecting the point with the minimum curvature value as the current seed point is:

[0088] Given a set of point clouds P = {P1, P2,..., P n}, coordinates {v1, v2,..., v n};

[0089] Calculate the mean vector i of the k neighborhood points of point P

[0090] Calculate point P according to the mean vector i The covariance matrix formed by the k neighborhood points of

[0091] Calculate the eigenvalues λ of the covariance matrix i1 ≤ λ i2 ≤ λ i3 and the corresponding eigenvectors e i1 、e i2 、e i3 ;

[0092] Obtain the curvature value according to the eigenvalues and the corresponding eigenvectors Select κ i The point with the smallest value as the initial seed point

[0093] Furthermore, in step (3), the geometric information extracted from the point cloud of the cylindrical component includes:

[0094] Use the PCA principal component analysis method to calculate the covariance eigenvector of the point cloud and obtain the center line direction of the cylindrical component;

[0095] Project the point cloud of the cylindrical component onto the cross-section XOY, and use the least squares method for point cloud fitting to obtain the center line position and radius of the cylindrical component;

[0096] Use the projection and grid division method to obtain the three-dimensional plane equations of the upper and lower end faces of the cylindrical component, including:

[0097] S1. Project the point cloud of the cylindrical component onto the longitudinal section XOZ, and obtain the boundary point cloud through the grid division method;

[0098] S2. Based on the boundary point cloud, use the RANSAC linear iterative fitting segmentation to obtain the two-dimensional linear equation of the projection of the cylindrical end face on the longitudinal section XOZ;

[0099] S3. According to the direction vector of the two-dimensional linear projection of the end face and the normal vector of the projection plane, obtain the normal vector of the end face, and then combine the coordinates of any point on the plane to obtain the three-dimensional plane equation of the end face.

[0100] Furthermore, projecting the point cloud of the cylindrical component onto the longitudinal section XOZ and obtaining the boundary point cloud through the grid division method includes:

[0101] Project the point cloud of the cylindrical component onto the longitudinal section XOZ plane and perform grid division to obtain the grid type, where the grid type includes: solid hole grid and empty hole grid;

[0102] Detect the grids adjacent to the current solid hole grid. If there is at least one empty hole grid, define the current solid hole grid as the boundary grid;

[0103] Obtain the boundary point cloud according to the boundary grid.

[0104] Furthermore, based on the boundary point cloud, using the RANSAC linear iterative fitting segmentation to obtain the projection equation of the cylinder end face point cloud on the two-dimensional plane includes:

[0105] S1. Select two points from the boundary point cloud to obtain the straight line equation;

[0106] S2. Calculate the distances from the remaining points in the boundary point cloud to the straight line equation, and count the number of inliers according to the distances;

[0107] S3. Based on the inliers, obtain the model parameters;

[0108] S4. Remove the inliers from the boundary point cloud to obtain a new point cloud;

[0109] S5. Repeat S1 - S4 until all the straight line equations, that is, the projection equations on the two-dimensional plane, are found.

[0110] Furthermore, extracting the geometric information of the hexahedron component point cloud includes:

[0111] Using the RANSAC plane fitting method to obtain the main plane equation of the hexahedron component point cloud, including:

[0112] S1. Randomly select three points from the main plane point cloud to obtain the plane equation;

[0113] S2. Calculate the distances from the remaining points in the boundary point cloud to the plane equation, and count the number of inliers according to the distances;

[0114] S3. Based on the inliers, obtain the model parameters;

[0115] S4. Remove the inliers from the boundary point cloud to obtain a new point cloud;

[0116] S5. Repeat S1 - S4 until all the plane equations, that is, the main plane equations, are found.

[0117] Use the projection and grid division method to obtain the side plane equation of the hexahedron component. The specific steps are similar to those for obtaining the cylinder end face equation;

[0118] Based on the main plane equation and the side plane equation, obtain the geometric information of the hexahedron component point cloud.

[0119] Furthermore, in step (4), for cylindrical components, it is necessary to model in their local coordinate systems, and for hexahedron components, model from points to solids. Function encapsulation is performed on the APDL modeling commands for each type of component, and the geometric parameters and position information obtained in step (3) are passed into the function. Finally, an APDL command file is output to generate a geometric finite element model in ANSYS software for subsequent analysis of the structure.

[0120] The present embodiment is described in detail below with reference to the accompanying drawings:

[0121] like Figure 5 The single-frame brick-wood structure shown in FIG. 1 is composed of a load-bearing wooden frame and a brick filling wall for enclosure. The example of the present invention provides a method for automatically generating a geometric finite element model of an ancient building structure based on a three-dimensional point cloud. The process is as follows: Figure 1 , which includes the following steps.

[0122] (1) Obtaining a point cloud model of the structure: Using a 3D laser scanner to scan the structure in all directions to obtain a 3D point cloud model of the structure;

[0123] Specifically, the structure is scanned in all directions using a Leica P40 3D laser scanner, and the scanned point cloud is registered to the global coordinate system using the software provided by the Leica P40 3D laser scanner. The outliers in the point cloud are filtered using a statistical filtering method, and voxel downsampling is performed to reduce the number of point clouds and improve computational efficiency. Finally, the PCA principal component analysis method is used to orderly search for a set of mutually orthogonal coordinate axes in the original space to ensure that the new coordinate axes can reflect the main change trend of the data to the greatest extent, and the three principal components of the structure point cloud are aligned with the X, Y, and Z axes of the global coordinate system, and the point cloud centroid is moved to the coordinate point O.

[0124] (2) Component segmentation: The surface of the component is separated by a regional segmentation algorithm. The shape type of the surface point cloud is determined by the relative relationship of the variance eigenvalues. The type of component to which the point cloud belongs is determined by the relative relationship of the size. Finally, the surface point cloud is grouped by the point cloud range and relative position relationship and the contact relationship is determined to obtain an independent component point cloud, such as Figure 2 shown.

[0125] Specifically, the region growing algorithm uses the angle between the curvature and the normal vector as a threshold to segment the point cloud. It can identify and segment areas where the curvature is discontinuous but the point cloud is continuous, thereby separating the surface point cloud of the component. The shape type of the surface point cloud can be determined based on the covariance eigenvalue. The change in one direction of the plane point cloud (usually the normal direction) is much smaller than the other two directions, and its corresponding covariance eigenvalue will be relatively small. The change in one direction of the cylindrical point cloud (usually along the axis of the cylinder) is much larger than the other two directions, and its corresponding covariance eigenvalue will be relatively large. Finally, the point cloud is grouped according to the point cloud range, relative position relationship, and size characteristics, the contact relationship is determined, and the building component category and number are marked.

[0126] (3) Geometric information extraction: For the point clouds of each component of the structure, the geometric information of the component is obtained through PCA principal component analysis, least squares method, grid division method, RANSAC point cloud fitting, etc. for modeling, such as Figure 4As shown, geometric information of various components is as Figure 3 shown.

[0127] Specifically, for cylindrical components, the centerline direction is obtained through PCA principal component analysis. The centerline position and cylindrical radius are obtained by projecting the point cloud onto the cross-section XOY and fitting the point cloud by the least squares method. Then, the point cloud is projected onto the longitudinal section XOZ, and the boundary point cloud is obtained through the mesh division method. The two-dimensional straight line equation of the projection of the cylindrical end face point cloud on the longitudinal section XOZ is obtained through RANSAC straight line iterative fitting segmentation. Then, the end face normal vector is obtained by combining the normal vector of the projection plane, and the three-dimensional plane equation of the end face is obtained by combining the coordinates of any point on the plane. For hexahedral components, first, the parametric equation of the main plane that can be collected by the point cloud is obtained through RANSAC plane fitting, and then the parametric equation of the hexahedral side is obtained by the same method as that of the cylindrical components. In addition, the contact surfaces between components are adjusted and merged to avoid abnormal modeling.

[0128] (4) Automatically output the geometric finite element model: Realize the parametric modeling, analysis, and visualization of the finite element model based on the ANSYS APDL language. Through the geometric and position information processed, analyzed, and identified in step (3), the APDL command file for parametric modeling of the components is automatically generated, and the geometric finite element model of the ancient building structure based on the three-dimensional point cloud is obtained.

[0129] Specifically, for cylindrical components, modeling needs to be carried out in their local coordinate systems. First, the coordinate system needs to be moved to the centerline position, and then the radius is input. For hexahedral components, the volume is generated from points for modeling. The APDL modeling commands for various types of components are encapsulated into functions, and the geometric parameters and position information obtained in (3) are passed into the functions. Finally, the APDL command file is output to generate the geometric finite element model in the ANSYS software, and subsequent analysis is carried out on the structure. The final generated model is as Figure 2 shown.

[0130] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An automatic generation method for the geometric finite element model of the ancient building structure based on 3D point cloud, characterized in that Including: Obtain the three-dimensional point cloud of the ancient building structure; Segment the three-dimensional point cloud to obtain independent component point clouds; Extract the geometric information of the component point clouds; Based on the geometric information, construct the geometric finite element model of the ancient building structure.

2. The automatic generation method of the geometric finite element model of the ancient building structure based on the three-dimensional point cloud according to claim 1, characterized in that, Segmenting the three-dimensional point cloud to obtain independent component point clouds includes: Use the region growing method to segment the surface point cloud of the component; Classify the surface point cloud of the component to obtain independent component point clouds, where the component point clouds include: cylindrical component point clouds and hexahedral component point clouds.

3. The automatic generation method of the geometric finite element model of the ancient building structure based on 3D point cloud according to claim 2, characterized in that, Using the region growing method to segment the surface point cloud of the component includes: S1. Set the number of points in the minimum point cluster, the number of points in the maximum point cluster, the smoothing threshold, and the curvature threshold; S2. Obtain the curvature values of the points in the surface point cloud of the component, sort the curvature values, and select the point with the minimum curvature value as the current seed point; S3. Add the current seed point to the seed point sequence and search for the neighborhood points of the current seed point; S4. Compare the angle between the normal of the neighborhood point and the normal of the current seed point, determine whether the angle satisfies the smoothing threshold, and add the current seed point that satisfies the smoothing threshold to the clustering region; S5. Calculate the curvature values of the neighborhood points, add the neighborhood points that satisfy the curvature threshold to the seed sequence, and update the current seed point; S6. Repeat S3 - S5 until the seed sequence is empty or the number of points in the point cluster generated by the remaining point cloud does not meet the number of points in the minimum point cluster.

4. The automatic generation method of the geometric finite element model of the ancient building structure based on 3D point cloud according to claim 3, characterized in that, The method of selecting the point with the minimum curvature value as the current seed point is: Given a set of point clouds P = {P1, P2, …, P n}; Calculate the mean vector of the k neighborhood points of point P i ; Calculate point P according to the mean vector i and the covariance matrix formed by the k neighborhood points of Calculate the eigenvalues and corresponding eigenvectors of the covariance matrix; Based on the eigenvalues and corresponding eigenvectors, obtain the curvature values.

5. The automatic generation method of the geometric finite element model of the ancient building structure based on 3D point cloud according to claim 2, characterized in that, Extracting the geometric information of the cylindrical component point cloud includes: Use the PCA principal component analysis method to calculate the covariance eigenvector of the point cloud to obtain the center line direction of the cylindrical component; Project the cylindrical component point cloud onto the cross-section XOY, and use the least squares method for point cloud fitting to obtain the center line position and radius of the cylindrical component; Based on the center line position and radius of the cylindrical component, use the projection and grid division method to obtain the three-dimensional plane equations of the upper and lower end faces of the cylindrical component.

6. The automatic generation method of the geometric finite element model of the ancient building structure based on 3D point cloud according to claim 5, characterized in that Using the projection and grid division method to obtain the three-dimensional plane equations of the upper and lower end faces of the cylindrical component includes: S1. Project the cylindrical component point cloud onto the longitudinal section XOZ, and obtain the boundary point cloud through the grid division method; S2. Based on the boundary point cloud, use the RANSAC line iterative fitting segmentation to obtain the projected two-dimensional line equation of the cylindrical end face on the longitudinal section XOZ; S3. Obtain the end face normal vector according to the end face projection two-dimensional line direction vector and the projection plane normal vector, and then combine with the coordinates of any point on the plane to obtain the three-dimensional plane equation of the end face.

7. A method for automatically generating a geometric finite element model of an ancient building structure based on 3D point clouds according to claim 6, characterized in that Projecting the cylindrical component point cloud onto the longitudinal section XOZ and obtaining the boundary point cloud through the grid division method includes: Project the cylindrical component point cloud onto the longitudinal section XOZ and perform grid division to obtain the grid types, where the grid types include: solid hole grids and empty hole grids; Detect the grids adjacent to the current solid hole grid, and if there is at least one empty hole grid, define the current solid hole grid as the boundary grid; Obtain the boundary point cloud according to the boundary grid.

8. The automatic generation method of the geometric finite element model of the ancient building structure based on the three-dimensional point cloud according to claim 6, characterized in that Based on the boundary point cloud, using the RANSAC linear iterative fitting segmentation to obtain the two-dimensional linear equation of the projection of the cylindrical end face point cloud on the longitudinal section XOZ includes: S1. Randomly select two points from the boundary point cloud to obtain the linear equation. S2. Calculate the distances from the remaining points in the boundary point cloud to the linear equation, and count the number of inliers according to the distances. S3. Based on the inliers, obtain the model parameters. S4. Remove the inliers from the boundary point cloud to obtain a new point cloud. S5. Repeat S1 - S4 until all the linear equations, that is, the two-dimensional linear equation, are found.

9. A method for automatically generating a geometric finite element model of an ancient building structure based on 3D point clouds according to claim 8, characterized in that Extract the geometric information of the hexahedron component point cloud, including: Using the RANSAC plane fitting method to obtain the main plane equation of the hexahedron component point cloud. Using the projection and grid division method to obtain the side plane equation of the hexahedron component. Based on the main plane equation and the side plane equation, obtain the geometric information of the hexahedron component point cloud. Among them, using the RANSAC plane fitting method to obtain the main plane equation of the hexahedron component point cloud includes: S1. Randomly select three points from the main plane point cloud to obtain the plane equation. S2. Calculate the distances from the remaining points in the boundary point cloud to the plane equation, and count the number of inliers according to the distances. S3. Based on the inliers, obtain the model parameters. S4. Remove the inliers from the boundary point cloud to obtain a new point cloud. S5. Repeat S1 - S4 until all the plane equations, that is, the main plane equation, are found. Using the projection and grid division method to obtain the side plane equation of the hexahedron component includes: Project the hexahedron component point cloud onto the main plane, and obtain the boundary point cloud through the grid division method. Based on the boundary point cloud, use the RANSAC linear iterative fitting segmentation to obtain the two-dimensional linear equation of the projection of the hexahedron side on the main plane. According to the end face projection two-dimensional linear direction vector and the projection plane normal vector, obtain the side normal vector, and then combine with the coordinates of any point on the plane to obtain the side plane equation.

10. A method for automatically generating a geometric finite element model of an ancient building structure based on three-dimensional point clouds according to claim 1, characterized in that Automatically output the geometric finite element model, and realize the parametric modeling, analysis, and visualization of the finite element model based on the ANSYS APDL language.

Images