A slicing method for constructing meshes from disordered 3D discrete point clouds

Through the disordered three-dimensional discrete point cloud slice networking method, the three-dimensional model construction process is simplified, the complex and time-consuming problem of existing algorithms is solved, and the three-dimensional geographic information data of any shape is quickly processed.

CN114463514BActive Publication Date: 2024-11-22WUHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210012797.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-01-07
Publication Date
2024-11-22
Estimated Expiration
2042-01-07

AI Technical Summary

Technical Problem

The existing three-dimensional networking algorithm is complex, time-consuming to process and high requirements for discrete points, which limits the rapid processing of three-dimensional geographic information data.

Method used

The disordered three-dimensional discrete point cloud slice networking method is adopted to construct a three-dimensional solid model through slice operation, convex hull and sorting combination, which simplifies the calculation process.

Benefits of technology

It realizes fast and simple three-dimensional model construction, fast processing speed, and is suitable for the rapid processing and real-time modeling of three-dimensional geographic information data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114463514B_ABST
    Figure CN114463514B_ABST
Patent Text Reader

Abstract

The present invention relates to a slicing meshing method for disordered three-dimensional discrete point clouds. First, a slicing operation is performed on disordered three-dimensional discrete points, and then the convex hull of the three-dimensional discrete point set is calculated on each slice, and finally the convex hull is combined in the order of slicing to complete the meshing of any three-dimensional entity. The slicing meshing algorithm proposed by the present invention is simple, has a fast processing speed, high calculation efficiency, has no requirements on data points, can process three-dimensional entities of any shape, and is particularly suitable for rapid processing of three-dimensional geographic information data and real-time three-dimensional modeling.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of three-dimensional geographic information processing, and in particular relates to a slicing network construction method for disordered three-dimensional discrete point clouds. Background Art

[0002] Unordered three-dimensional discrete point clouds are the basic data source for three-dimensional spatial information processing. In order to achieve effective expression and analysis of three-dimensional spatial information, three-dimensional discrete point clouds are usually fitted into three-dimensional entities to form specific geometric targets. For example, unordered three-dimensional discrete point clouds can be used to form a network to form terrain data, building models, highway models, overpass models, railway models, etc. However, how to simply and quickly build a three-dimensional model has always been a hot topic in the industry. At present, the most common algorithm is the Poisson three-dimensional model network method, referred to as the Poisson network method. This method has rigorous theory and good versatility, but the calculation is quite complicated. During the calculation process, the normal vectors of all discrete points need to be known. Each expansion forms a new normal vector, and the algorithm has poor anti-interference ability. In addition to the Poisson network method, the two-dimensional Delaunay triangulation plus depth attributes can also be used to complete the three-dimensional model network, referred to as the Delaunay method. The Delaunay method is also a relatively mature algorithm and is widely used in geographic information processing, but it cannot express data models with multiple coverage in the depth direction. It is generally only used to describe terrain features and cannot describe complex three-dimensional entities. Another relatively mature 3D meshing algorithm is to construct a tetrahedron first, then use the graph cut algorithm to optimize the tetrahedron, and finally generate a 3D model, which is referred to as the tetrahedron method. The tetrahedron method can describe 3D models of any shape and has good anti-interference performance, but the algorithm is relatively complex, especially the optimization process requires repeated iterations, which is very time-consuming. In addition, the tetrahedron method, like the Poisson meshing method, also requires the normal vectors of all discrete points. In summary, the existing 3D meshing algorithms are relatively complex, very time-consuming to process, and have high requirements for the discrete points of the meshing, which limits the rapid processing of 3D geographic information data. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention provides a slicing method for constructing a network of disordered three-dimensional discrete point clouds. First, a slicing operation is performed on the disordered three-dimensional discrete points, and then the convex hull of the three-dimensional discrete point set is calculated on each slice, and finally the convex hulls are combined in the order of slicing to complete the network construction of any three-dimensional entity.

[0004] In order to achieve the above object, the technical solution provided by the present invention is a slicing method for constructing a network of disordered three-dimensional discrete point clouds, comprising the following steps:

[0005] Step 1, obtaining the effective range and density of disordered three-dimensional discrete points;

[0006] Step 1.1, count the maximum and minimum coordinates of the three-dimensional discrete points in the X, Y, and Z directions, use the histogram statistical algorithm to remove the noise points in the point cloud, and obtain the effective range of the point cloud;

[0007] Step 1.2, using the existing point cloud density calculation algorithm to obtain the point cloud density;

[0008] Step 2, select the direction with the largest effective range in the three directions of X, Y, and Z as the main direction, and then slice the disordered three-dimensional discrete points along the main direction at intervals of the density value of the direction;

[0009] Step 3, find the convex hull of the discrete point set on the slice surface of all slice data in order;

[0010] Step 4, sort all the convex hulls of the cut surfaces according to the main direction, and construct a triangulated network within the closed loop for the first cut surface and the last cut surface, that is, construct the two cut surfaces as closed surfaces;

[0011] Step 4.1, find a point in the convex hull on the section that is closest to the centroid as the first point to be added to the closed loop, and connect it with each point on the closed loop to form a triangulated network after the first insertion;

[0012] Step 4.2, based on the current triangulated network, determine which triangle the remaining points fall into, add the points and connect them with the three vertices of this triangle to form three triangles;

[0013] Step 4.3, repeat step 4.2 until all points are added, and the triangulated network can be constructed in the closed loop;

[0014] Step 5: triangulate the boundaries of two adjacent convex hulls in the order of the main directions, and finally complete the surface meshing of all point clouds;

[0015] Step 5.1: For each vertex of the left convex hull, find the vertex closest to it on the right convex hull and connect them. It is allowed that multiple points correspond to one point, and it is also allowed that one point corresponds to multiple points.

[0016] Step 5.2: For the vertices that are not connected in the right convex hull, search for the nearest vertices in the left convex hull for connection, ensuring that all vertices of the two adjacent convex hulls are connected;

[0017] Step 5.3, determine whether all connected surfaces form a triangulated network. If not, triangulate the remaining sub-blocks until a triangulated network is formed between two adjacent convex hulls.

[0018] Step 5.4, perform the meshing operation of step 5.3 on all slices along the main direction sequence, and finally complete the meshing of the point cloud surface.

[0019] Moreover, in step 1.1, the maximum and minimum values ​​of the point cloud coordinates in the three directions of X, Y, and Z are first counted. , , To obtain the effective range of the point cloud, the histograms of the points in the three directions of X, Y, and Z are counted respectively. Then, the noise points at the maximum and minimum boundaries of the three histograms are deleted. , , The calculation formula is as follows:

[0020] (1)

[0021] In the formula, , , , , , is the maximum and minimum value in the X, Y, and Z directions, and n is the histogram Figure 1 The dimension of the array.

[0022] Moreover, the convex hull of each discrete point set of the slice is obtained in step 3, that is, it is converted into a convex hull problem in a two-dimensional plane, and a section plane coordinate system is established with two coordinate axes perpendicular to the main direction, and then the convex hull of each section is obtained by using the classical wrapping method; first, the point with the smallest y coordinate in the section discrete point set is selected as the first point of the convex hull, and if there are multiple points with the same y coordinate and they are all minimum values, then the point with the smallest z coordinate is selected as the first point of the convex hull; then, with the first point as the endpoint, a ray is drawn in the opposite direction of the y coordinate and rotated counterclockwise, and when the ray passes through the first point in the discrete point set, this point is used as the second point of the convex hull; then, with the current ray direction as the reference and the current point as the endpoint, continue to rotate counterclockwise to find the next point in the discrete point set that passes through the ray, and set this point as the current point, and finally repeat this process until you return to the first point, and the line connecting all the endpoints obtained at this time is the desired convex hull.

[0023] Moreover, in the step 5.3, the connecting line between the two convex hulls is used as the basic judgment unit, and the connecting line of each vertex is traversed counterclockwise along the left convex hull, and its endpoint on the left convex hull is point 1, and its endpoint on the right convex hull is point 2. If there is a point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points 1 and 2 of the connecting line, then these three points have formed a triangle, and the judgment of the next connecting line continues; if there is no point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points of the connecting line, then the current vertex in the left convex hull is used. Take the endpoint of the connecting line as the starting point, search for the next point in the counterclockwise direction as point 3, and at the same time, take the endpoint of the current connecting line in the right convex hull as the starting point and search for the next point in the counterclockwise direction as point 4. At this time, points 1 and 2 are connected, and points 1 and 3 are both located in the left convex hull, and points 2 and 4 are both located in the right convex hull. Select points 3 and 2, and points 4 and 1 with shorter distances to connect and obtain new connecting lines; repeat the above operations for the newly generated connecting lines until no new connecting lines are generated, that is, the triangulation network construction between two adjacent convex hulls is completed.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] The slice meshing algorithm is simple, with fast processing speed, high computational efficiency, and no requirements for data points. It can process three-dimensional entities of any shape and is particularly suitable for rapid processing of three-dimensional geographic information data and real-time three-dimensional modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of an embodiment of the present invention.

[0027] Figure 2 Schematic diagram of the principle of removing point cloud noise using the histogram statistical algorithm of the present invention.

[0028] Figure 3 It is a schematic diagram of the slicing principle of the present invention.

[0029] Figure 4 This is a schematic diagram of obtaining the convex hull of discrete points using the wrapping method of the present invention.

[0030] Figure 5 It is a schematic diagram of the internal segmentation network construction principle of two sections at both ends of the present invention.

[0031] Figure 6 The figure is a schematic diagram of the principle of finding the nearest point of the convex hull endpoints of any two cut surfaces of the present invention.

[0032] Figure 7 This is an example of the network construction result of this method. DETAILED DESCRIPTION

[0033] The present invention provides a slicing method for constructing a network of disordered three-dimensional discrete point clouds. The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0034] like Figure 1 As shown, the process of the embodiment of the present invention includes the following steps:

[0035] Step 1: Obtain the effective range and density of disordered three-dimensional discrete points.

[0036] Step 1.1, count the maximum and minimum coordinates of the three-dimensional discrete points in the X, Y, and Z directions, use the histogram statistical algorithm to remove noise points in the point cloud, and obtain the effective range of the point cloud.

[0037] First, count the maximum and minimum values ​​of the point cloud coordinates in the three directions of X, Y, and Z. , , To calculate the interval, the number of points in the three directions of X, Y, and Z is counted in histograms, and then the noise points at the maximum and minimum boundaries of the three histograms are deleted (such as Figure 2 As shown, delete and All points within the coordinate range are deleted) to obtain the valid range of the point cloud.

[0038] , , The calculation formula is as follows:

[0039] (1)

[0040] In the formula, , , , , , is the maximum and minimum value in the X, Y, and Z directions, and n is the histogram Figure 1 The dimension of the dimensional array is 1000 in this embodiment.

[0041] Step 1.2, use existing algorithms such as octree statistics method and grid statistics method to calculate the point cloud density.

[0042] The point cloud density is defined as the mean distance between any two adjacent points in the point cloud, which is described by three components dx, dy, and dz. The point cloud density can be calculated using existing algorithms such as octree statistics and grid statistics.

[0043] Step 2: Select the direction with the largest effective range (such as X) as the main direction in the three directions of X, Y, and Z, and then slice the disordered three-dimensional discrete points along the main direction with the density value of this direction (such as dx) as the interval. The slicing principle is as follows: Figure 3 shown.

[0044] Step 3, find the convex hull of the discrete point set on the slice surface of all slice data in sequence.

[0045] Find the convex hull of each discrete point set of the slice, that is, convert it into the problem of finding the convex hull in a two-dimensional plane, establish the section plane coordinate system with two coordinate axes perpendicular to the main direction (such as Y and Z axes), and then use the classic wrapping method to find the convex hull of each section.

[0046] like Figure 4 As shown, first select the point with the smallest y coordinate in the discrete point set of the section (if there are multiple points with the same y coordinate and they are all minimum values, then select the point with the smallest z coordinate) as the first point of the convex hull; then use the first point as the endpoint, draw a ray in the opposite direction of the y coordinate and rotate counterclockwise. When the ray passes through the first point in the discrete point set, use this point as the second point of the convex hull; then use the current ray direction as the reference and the current point as the endpoint, continue to rotate counterclockwise to find the next point in the discrete point set that passes through the ray, and set this point as the current point. Finally, repeat this process until you return to the first point. At this time, the line connecting all the endpoints obtained is the desired convex hull.

[0047] Step 4: Sort all the convex hulls of the cut surfaces by the main direction (such as the X direction), and construct a triangulated network within the closed loop for the first and last cut surfaces, that is, construct the two cut surfaces as closed surfaces. The network construction principle is as follows: Figure 5 shown.

[0048] Step 4.1, find a point in the convex hull on the section that is closest to the centroid as the first point to be added to the closed loop, and connect it with each point on the closed loop to form a triangulated network after the first insertion.

[0049] Step 4.2, based on the current triangulated network, determine which triangle the remaining points fall into, add the points and connect them with the three vertices of this triangle to form three triangles.

[0050] Step 4.3, repeat step 4.2 until all points are added, and the triangulated network can be constructed in a closed loop.

[0051] Step 5: Triangulate the boundaries of two adjacent convex hulls in the order of the main direction (such as the x direction), and finally complete the surface mesh of all point clouds. The meshing principle is as follows: Figure 6 shown.

[0052] Step 5.1: For each vertex of the left convex hull, find the vertex closest to it on the right convex hull (i.e., the convex hull of the next cut surface) and connect them. It is allowed that multiple points correspond to one point, and it is also allowed that one point corresponds to multiple points.

[0053] Step 5.2: For the vertices that are not connected in the right convex hull, search for the nearest vertices in the left convex hull for connection, ensuring that all vertices of the two adjacent convex hulls are connected.

[0054] Step 5.3, determine whether all connected surfaces form a triangulated network. If not, triangulate the remaining sub-blocks until a triangulated network is formed between two adjacent convex hulls.

[0055] Take the connecting line between the two convex hulls as the basic judgment unit, traverse the connecting line of each vertex counterclockwise along the left convex hull, and its endpoint on the left convex hull is point 1, and its endpoint on the right convex hull is point 2. If there is a point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points of the connecting line, points 1 and 2, then these three points have formed a triangle, and continue to judge the next connecting line; if there is no point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points of the connecting line, then take the endpoint of the current connecting line in the left convex hull as the starting point, search for the next point in the counterclockwise direction as point 3, and at the same time, take the endpoint of the current connecting line in the right convex hull as the starting point, search for the next point in the counterclockwise direction as point 4. At this time, points 1 and 2 are connected, and points 1 and 3 are both located in the left convex hull, and points 2 and 4 are both located in the right convex hull. Select points 3 and 2, and points 4 and 1 with a shorter distance to connect, and get a new connecting line. Repeat the above operation for the newly generated connecting lines until no new connecting lines are generated, that is, the triangulated network construction process between two adjacent convex hulls is completed.

[0056] by Figure 6 Take (c) as an example. For the current connecting line AG, there is already a point B in the counterclockwise direction that forms a triangle with AG, so the next connecting line BG is determined. For the BG connecting line, there is no point in the counterclockwise direction of the current left and right convex hulls that forms a triangle with it. Then compare the size of BH and GC. Since BH>GC, point C is selected to connect and form a triangle. Then perform the above judgment operation on the newly generated connecting line CG until no new connecting line is generated, thereby completing the triangulation network processing between two adjacent convex hulls.

[0057] Step 5.4, perform the meshing operation of step 5.3 on all slices along the main direction sequence, and finally complete the meshing of the point cloud surface.

[0058] During specific implementation, the above process can be automatically operated using computer software technology.

[0059] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A slicing method for constructing a network of disordered three-dimensional discrete point clouds, characterized in that: The following steps are involved: Step 1, obtaining the effective range and density of disordered three-dimensional discrete points; Step 1.1, count the three-dimensional discrete points X, Y, Z The maximum and minimum coordinates of the direction are calculated, and the noise points in the point cloud are removed using the histogram statistical algorithm to obtain the effective range of the point cloud; Step 1.2, using the existing point cloud density calculation algorithm to obtain the point cloud density; Step 2, X, Y, Z The direction with the largest effective range is selected as the main direction among the three directions, and then the disordered three-dimensional discrete points are sliced ​​along the main direction at intervals of the density value of the direction; Step 3, find the convex hull of the discrete point set on the slice surface of all slice data in order; Step 4, sort all the convex hulls of the cut surfaces according to the main direction, and construct a triangulated network within the closed loop for the first cut surface and the last cut surface, that is, construct the two cut surfaces as closed surfaces; Step 5: triangulate the boundaries of two adjacent convex hulls in the order of the main directions, and finally complete the surface meshing of all point clouds; Step 5.1: For each vertex of the left convex hull, find the vertex closest to it on the right convex hull and connect them. It is allowed that multiple points correspond to one point, and it is also allowed that one point corresponds to multiple points. Step 5.2: For the vertices that are not connected in the right convex hull, search for the nearest vertices in the left convex hull to connect them, ensuring that all vertices of the two adjacent convex hulls are connected; Step 5.3, determine whether all connected surfaces form a triangulated network. If not, triangulate the remaining sub-blocks until a triangulated network is formed between two adjacent convex hulls. Take the connecting line between the two convex hulls as the basic judgment unit, traverse the connecting line of each vertex counterclockwise along the left convex hull, its endpoint on the left convex hull is point 1, and its endpoint on the right convex hull is point 2. If there is a point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points 1 and 2 of the connecting line, then these three points have formed a triangle, and continue to judge the next connecting line; if there is no point in the counterclockwise direction of the current vertex of the left and right convex hulls that is connected to both the two end points of the connecting line, then the end point of the current connecting line in the left convex hull is used. Take point 1 as the starting point, search for the next point in the counterclockwise direction as point 3, and at the same time, take the endpoint of the current connection line in the right convex hull as the starting point and search for the next point in the counterclockwise direction as point 4. At this time, points 1 and 2 are connected, and points 1 and 3 are both located in the left convex hull, and points 2 and 4 are both located in the right convex hull. Select points 3 and 2, and points 4 and 1 with shorter distances to connect them to obtain new connection lines; repeat the above operations for the newly generated connection lines until no new connection lines are generated, that is, the triangulation network construction between two adjacent convex hulls is completed; Step 5.4, perform the meshing operation of step 5.3 on all slices along the main direction sequence, and finally complete the meshing of the point cloud surface.

2. A method for constructing a slicing network of an unordered three-dimensional discrete point cloud according to claim 1, characterized in that: In step 1.1, we first count X, Y, Z The maximum and minimum values ​​of the point cloud coordinates in three directions are , , For interval X, Y, Z The number of points in the three directions is counted in histograms respectively, and then the noise points at the maximum and minimum boundaries of the three histograms are deleted to obtain the effective range of the point cloud; , , The calculation formula is as follows: (1) In the formula, , , , , , for X, Y, Z The maximum and minimum values ​​in the direction, n is the dimension of the one-dimensional array of the histogram.

3. The method for constructing a slicing network of an unordered three-dimensional discrete point cloud according to claim 1, characterized in that: In step 3, the convex hull of each discrete point set of the slice is calculated, that is, it is converted into a convex hull problem in a two-dimensional plane. The section plane coordinate system is established with two coordinate axes perpendicular to the main direction, and then the convex hull of each section is calculated using the classic wrapping method. First, the point with the smallest y coordinate in the section discrete point set is selected as the first point of the convex hull. If there are multiple points with the same y coordinate and they are all minimum values, then the point with the smallest z coordinate is selected as the first point of the convex hull. Then, with the first point as the endpoint, a ray is drawn in the opposite direction of the y coordinate and rotated counterclockwise. When the ray passes through the first point in the discrete point set, this point is used as the second point of the convex hull. Then, with the current ray direction as the reference and the current point as the endpoint, continue to rotate counterclockwise to find the next point in the discrete point set that passes through the ray, and set this point as the current point. Finally, this process is repeated until you return to the first point. The line connecting all the endpoints obtained at this time is the desired convex hull.

4. The method for constructing a slicing network of an unordered three-dimensional discrete point cloud according to claim 1, characterized in that: Step 4 includes the following sub-steps: Step 4.1, find a point in the convex hull on the section that is closest to the centroid as the first point to be added to the closed loop, and connect it with each point on the closed loop to form a triangulated network after the first insertion; Step 4.2, based on the current triangulated network, determine which triangle the remaining points fall into, add the points and connect them with the three vertices of this triangle to form three triangles; Step 4.3, repeat step 4.2 until all points are added, and the triangulated network can be constructed in a closed loop.