A method for registering multi-temporal laser point clouds with three-dimensional real scene models

Abstract

The application provides a multi-time-phase laser point cloud and three-dimensional real scene model registration method, and belongs to the technical field of point cloud registration, and comprises the following steps: sequentially sorting and automatically registering multi-period laser radar point cloud data in a long time period, and calculating point cloud density values. The initial vertex point cloud of the three-dimensional real scene model is extracted, three-dimensional grid creation is performed according to the calculated density values, the curvature adaptive sampling method and the grid plane interpolation method are adopted to expand the point cloud density, and density-enhanced point cloud data is obtained. Feature point information extraction and matching are performed on the point cloud data obtained by automatically registering the multi-time-phase point cloud and the density-enhanced point cloud data, and the automatic registration of the two types of point clouds is realized. By adjusting the calculation model point cloud to meet the target density and be uniformly distributed, the problem of registration difficulty caused by the large density difference between the three-dimensional real scene model point cloud and the multi-time-phase point cloud is effectively solved.

Description

Technical Field

[0001] This invention belongs to the field of point cloud registration technology, specifically relating to a registration method for multi-temporal laser point clouds and three-dimensional real scene models. Background Technology

[0002] Ancient buildings bear witness to the history of ancient humans and are important cultural relics. In the early stages of cultural relic protection, investigations into damage to cultural relics mainly focused on recording and summarizing the types, distribution, causes, and severity of damage. However, such investigations could not delve into the subtle changes in damage. Previously, close-range photogrammetry was often used to detect damage to cultural relics, and the plan and elevation maps generated by close-range photogrammetry could achieve millimeter-level accuracy. However, climatic conditions, shooting angle, and lighting conditions significantly affect the results of close-range photogrammetry. Therefore, this method is not entirely effective.

[0003] The emergence of 3D laser scanning technology has provided a new method for detecting damage to cultural relics. The most significant advantage of this technology lies in its ability to directly acquire high-density 3D point cloud data, thereby reflecting the details of the surface of the object being measured. In the construction and management of cultural relics, the registration between lidar point clouds and 3D reality models can be used to detect differences between the design model and the actual state of the ancient building captured in the point cloud. The main challenge is registering the point cloud extracted from the 3D reality model with existing point clouds. Many applications require cross-format registration of different types of 3D spatial data (such as point clouds, meshes, and Building Information Modeling (BIM)).

[0004] In the field of automatic registration of multi-temporal laser point clouds and 3D reality models, density inconsistencies easily arise between point cloud data from different sources during the fusion and registration process. Existing technologies improve density inconsistencies by interpolating sparse point clouds, but when interpolating point clouds into a mesh plane to expand density, a balance between feature information and reasonable distribution cannot be achieved. This results in poor correspondence between laser point cloud information and 3D reality point cloud positions, and the final fused point cloud suffers from low accuracy and large deformation monitoring errors. Summary of the Invention

[0005] To address the shortcomings of existing methods for registering laser point clouds with 3D reality models, this invention provides a registration method for multi-temporal laser point clouds with 3D reality models.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A registration method for multi-temporal laser point clouds and 3D reality models includes the following steps: Multiple phases of lidar point cloud data of ancient buildings within a set time period are acquired. Feature points of the point cloud data between two adjacent phases are extracted sequentially according to the time sequence, and feature point matching is performed to obtain multi-temporal point cloud data. The point cloud density value is calculated based on the multi-temporal point cloud data. Vertex point cloud is extracted from the 3D real scene model to obtain initial vertex point cloud data; using the point cloud density value as a standard, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data. The multi-temporal point cloud data and density-enhanced point cloud data are automatically registered, and the deformation monitoring of ancient buildings within a set time period is realized based on the registration results.

[0007] Preferably, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data, including the following steps: A 3D mesh model is created based on the vertex point cloud, and the curvature at the vertex of each triangular facet is calculated using the curvature adaptive sampling method. The curvature of all triangular facets is normalized. Based on the normalized facet curvature value and point cloud density value, the target point cloud sampling density of each triangular facet is defined. The target point cloud sampling density of each triangular facet is then mapped to the target side length. A judgment rule is set between the maximum side length and the target side length of the original triangular facet. Based on the preset rule, it is determined whether the triangular facet needs to be subdivided until all triangular facets meet the target side length or reach the maximum number of subdivision layers. Centroid interpolation sampling is performed on each subdivided triangular face; the minimum point distance is determined based on the surface sampling density mapping, and the interpolation points are filtered by variable radius Poisson filtering to remove overly dense points, resulting in density-enhanced point cloud data.

[0008] Preferably, the curvature adaptive sampling method is used to calculate the curvature at each vertex of the triangular facet, specifically through the following formula: ; in, , and These represent the curvature of each vertex of each triangular facet.

[0009] Preferably, the preset rule is specifically to calculate the current maximum side length of each original triangle face. If the current maximum side length is greater than the target side length, the face needs to be subdivided. For the triangle face that needs to be subdivided, a new vertex is inserted at the midpoint of the three sides to divide the triangle face into four sub-triangle faces. The sub-triangle faces inherit the curvature value of the parent face. All triangle faces are subdivided recursively.

[0010] Preferably, the centroid interpolation sampling on each subdivided triangular face is specifically performed by determining the number of sampling points on each subdivided triangular face based on the face area and the original curvature mapping, and then performing centroid interpolation, normal interpolation and attribute interpolation on each sampling point.

[0011] Preferably, the multi-temporal point cloud data and the density-enhanced point cloud data are automatically registered using the point cloud tool cloud compare.

[0012] This invention also provides a registration system for multi-temporal laser point clouds and 3D reality models, specifically including: The first data module is used to acquire multi-phase lidar point cloud data of ancient buildings within a set time period, extract feature points between adjacent phases of point cloud data in sequence according to the time series, and perform feature point matching to obtain multi-temporal point cloud data; and calculate the point cloud density value based on the multi-temporal point cloud data.

[0013] The second data module is used to extract vertex point clouds from the 3D real-world model to obtain initial vertex point cloud data; using the point cloud density value as a standard, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data.

[0014] The registration module is used to automatically register the multi-temporal point cloud data and the density-enhanced point cloud data, and to monitor the deformation of ancient buildings within a set time period based on the registration results.

[0015] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps described in the method for registering multi-temporal laser point clouds with a three-dimensional real scene model.

[0016] The present invention also provides a computer-readable storage medium storing a computer program, which, when loaded by a processor, is capable of executing the steps described in the method for registering multi-temporal laser point clouds with a three-dimensional real-world model.

[0017] The registration method between multi-temporal laser point clouds and 3D reality models provided by this invention has the following beneficial effects: This invention first completes adaptive registration of multi-temporal lidar point clouds, and then achieves cross-format registration between the 3D real-scene model point cloud and the registration result. This provides a high-precision data foundation for subsequent comparison of differences between the design model and the actual scanning state, and is particularly suitable for long-term monitoring of minute deformations and damage changes in ancient buildings. The point cloud density value calculated from the multi-temporal point cloud data is used as a standard. Curvature adaptive sampling and grid plane interpolation methods are used to densify the initial vertex point cloud data, ensuring consistent point cloud density during the registration process and reducing the defects caused by uneven point cloud distribution leading to poor registration results. The multi-temporal point cloud data and the density-enhanced point cloud data are automatically registered to achieve deformation monitoring of ancient buildings within a set time period. This provides an efficient and reliable technical means for the refined and automated detection of damage to ancient architectural relics. Attached Figure Description

[0018] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a registration method between multi-temporal laser point clouds and a 3D real-world model, as described in an embodiment of the present invention.

[0020] Figure 2 This is a flowchart for extracting high-density point cloud data from a 3D reality model in an embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram of registration between multiple temporal point clouds in an embodiment of the present invention.

[0022] Figure 4 This is a schematic diagram of fixed-point cloud curvature calculation in an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0024] Example This invention provides a method for registering multi-temporal laser point clouds with a 3D reality model, such as... Figure 1 As shown, the specific steps include: Step 1: Acquire multi-phase LiDAR point cloud data of ancient buildings over a long period. Use Cloud Compare to extract feature points between two phases of point cloud data, sorted by time series, and perform feature point matching to achieve automatic registration between multi-temporal point clouds. Figure 3 As shown, duplicate point cloud data are removed from the new point cloud obtained by multi-temporal automatic registration, and the point cloud density value is directly calculated using point cloud density calculation tools (such as Cloud Compare, Global Mapper, etc.).

[0025] Step 2: Use software such as Cloud Compare to extract vertex point clouds from the 3D reality model to obtain initial vertex point clouds. At this point, the data is relatively sparse, and the density values ​​differ significantly from those obtained by multi-temporal point cloud registration.

[0026] Step 3: Create a 3D mesh from the initial vertex point cloud. Using the point cloud density value obtained in Step 1 as the standard, use the mesh plane interpolation method to increase the point cloud density and obtain high-density point cloud data.

[0027] The specific processing flow for extracting high-density point cloud data from 3D reality models, such as... Figure 2 As shown, the specific steps include: (1) Calculate the curvature of the triangular surface, such as Figure 4 As shown, a 3D mesh model is created based on the vertex point cloud. The quality of the triangular mesh varies, and an adaptive curvature sampling method is used for each triangular face. First, calculate the curvature at the vertex of each triangular facet. Specifically, obtain the face curvature through interpolation. The interpolation value is:

[0028] ; in, , and These represent the curvature of each vertex of each triangular facet.

[0029] (2) Perform surface sampling density mapping. First, apply the formula to all triangular patches. Curvature normalization is performed, and then the target point cloud sampling density on each triangular face is defined through a surface sampling density mapping:

[0030] ; in, This represents the sampling density of the target point cloud on the triangular face. Given the known target point cloud density in step one, The normalized surface curvature value. The mapping function from curvature to density is the mapping weight. The total number of points is distributed according to the curvature weight multiplied by the area, achieving both local adaptation and global conservation (fixed total number of points). Since mesh subdivision is a geometric operation, it can only be driven by geometric quantities (edge ​​lengths), while point density is a statistical quantity and cannot be directly used for subdivision judgment. Therefore, the target density needs to be converted into the target edge length.

[0031] The subdivided mesh is mapped to the target side length based on its density, enabling the determination of whether each triangular face needs further subdivision based on its specific side length. For an approximately equilateral triangle, the target side length and the target point density satisfy the following equation:

[0032] ; That is, the target side length is inversely proportional to the square root of the target point density. The target side length of the f-th triangular face is calculated by the following formula: ; in, For the minimum allowable side length, This represents the maximum permissible side length. It enables high curvature, small side length, and multiple subdivisions, as well as low curvature, large side length, and fewer subdivisions.

[0033] (3) Curvature-adaptive triangular mesh subdivision. For each original triangular face, calculate its current maximum side length. Determine whether the triangular face needs to be subdivided. If it is larger than the side length of the target sampling patch, then... Then the face needs to be subdivided. For the triangle that needs to be subdivided, insert a new vertex at the midpoint of the three edges to divide the triangle into four sub-triangles. The sub-triangles inherit the curvature value of the parent face. Repeat the comparison between the maximum side length and the target side length until all triangles meet the target side length or the maximum number of subdivision levels is reached.

[0034] (4) Perform centroid interpolation sampling on each triangular face. Determine the number of sampling points on each subdivided triangular face:

[0035] ; in, For surface area, The original curvature is mapped, and then centroid interpolation, normal phase interpolation, and attribute interpolation are performed on each sampling point.

[0036] (5) Based on the above surface sampling density mapping, determine the minimum distance at the point level, and use software such as Cloud Compare to perform variable radius Poisson filtering on each interpolation point, retaining points with a distance greater than the minimum distance and removing oversampled points. This ensures that the final point cloud meets the target density and is uniformly distributed.

[0037] Step 4: In existing software such as Cloud Compare, feature point information is extracted and matched between the point cloud data obtained from the automatic registration of multi-temporal point clouds (multi-temporal point clouds) and the interpolated high-density point cloud data (3D reality model), achieving automatic registration of the two types of point clouds. Ancient buildings have undergone many years of weathering and changes; by registering and fusing data from multiple periods, the changes and damage to ancient buildings over the years can be visually visualized.

[0038] This invention also provides a registration system for multi-temporal laser point clouds and three-dimensional reality models, comprising: The first data module is used to acquire multi-phase lidar point cloud data of ancient buildings within a set time period, extract feature points between adjacent phases of point cloud data in sequence according to the time series, and perform feature point matching to obtain multi-temporal point cloud data; and calculate the point cloud density value based on the multi-temporal point cloud data.

[0039] The second data module is used to extract vertex point clouds from the 3D real-world model to obtain initial vertex point cloud data. Using the point cloud density value as a standard, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data.

[0040] The registration module is used to automatically register multi-temporal point cloud data and density-enhanced point cloud data, and to monitor the deformation of ancient buildings within a set time period based on the registration results.

[0041] The modules in the aforementioned multi-temporal laser point cloud and 3D reality model registration system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0042] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in an embodiment of a method for registering multi-temporal laser point clouds with a 3D real-world model. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0043] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions that can be executed by a processor of a computer device to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in the embodiment of a method for registering multi-temporal laser point clouds with a 3D real-world model. Specific implementation methods can be found in the method embodiments, which will not be repeated here.

[0044] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0048] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the present invention patent. No reference numerals in the claims should be construed as limiting the scope of the claims. Any simple variations or equivalent substitutions of technical solutions that can be readily obtained by those skilled in the art within the scope of the technology disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method for registering multi-temporal laser point clouds with a 3D real-world model, characterized in that, Includes the following steps: Multiple phases of lidar point cloud data of ancient buildings within a set time period are acquired. Feature points of the point cloud data between two adjacent phases are extracted sequentially according to the time sequence, and feature point matching is performed to obtain multi-temporal point cloud data. The point cloud density value is calculated based on the multi-temporal point cloud data. Vertex point cloud is extracted from the 3D real scene model to obtain initial vertex point cloud data; using the point cloud density value as a standard, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data. The multi-temporal point cloud data and density-enhanced point cloud data are automatically registered, and the deformation monitoring of ancient buildings within a set time period is realized based on the registration results.

2. The registration method for multi-temporal laser point clouds and three-dimensional real-scene models according to claim 1, characterized in that, The initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data, including the following steps: A 3D mesh model is created based on the vertex point cloud, and the curvature at the vertex of each triangular facet is calculated using the curvature adaptive sampling method. The curvature of all triangular facets is normalized. Based on the normalized facet curvature value and point cloud density value, the target point cloud sampling density of each triangular facet is defined. The target point cloud sampling density of each triangular facet is then mapped to the target side length. A judgment rule is set between the maximum side length and the target side length of the original triangular facet. Based on the preset rule, it is determined whether the triangular facet needs to be subdivided until all triangular facets meet the target side length or reach the maximum number of subdivision layers. Centroid interpolation sampling is performed on each subdivided triangular face; the minimum point distance is determined based on the surface sampling density mapping, and the interpolation points are filtered by variable radius Poisson filtering to remove overly dense points, resulting in density-enhanced point cloud data.

3. The registration method for multi-temporal laser point clouds and three-dimensional real-scene models according to claim 2, characterized in that, The curvature adaptive sampling method is used to calculate the curvature at the vertex of each triangular facet, specifically through the following formula: ； in, , and These represent the curvature of each vertex of each triangular facet.

4. The registration method for multi-temporal laser point clouds and three-dimensional real-scene models according to claim 2, characterized in that, The preset rule is as follows: for each original triangle, calculate its current maximum side length. If the current maximum side length is greater than the target side length, the face needs to be subdivided. For the triangle that needs to be subdivided, insert a new vertex at the midpoint of the three sides to divide the triangle into four sub-triangles. The sub-triangles inherit the curvature value of the parent face. Recursively subdivide all triangles.

5. The registration method for multi-temporal laser point clouds and three-dimensional real-scene models according to claim 2, characterized in that, The centroid interpolation sampling is performed on each subdivided triangular face. Specifically, for each subdivided triangular face, the number of sampling points on the face is determined based on the face area and the original curvature mapping. Then, centroid interpolation, normal interpolation, and attribute interpolation are performed on each sampling point.

6. The registration method for multi-temporal laser point clouds and three-dimensional real-scene models according to claim 1, characterized in that, The multi-temporal point cloud data and density-enhanced point cloud data were automatically registered using the point cloud tool Cloud Compare.

7. A registration system for multi-temporal laser point clouds and 3D real-scene models, characterized in that, include: The first data module is used to acquire multi-phase lidar point cloud data of ancient buildings within a set time period, extract feature points between adjacent phases of point cloud data in a time sequence, and perform feature point matching to obtain multi-temporal point cloud data; and calculate the point cloud density value based on the multi-temporal point cloud data. The second data module is used to extract vertex point clouds from the 3D real scene model to obtain initial vertex point cloud data; using the point cloud density value as a standard, the initial vertex point cloud data is encrypted using curvature adaptive sampling and mesh plane interpolation methods to obtain density-enhanced point cloud data. The registration module is used to automatically register the multi-temporal point cloud data and the density-enhanced point cloud data, and to monitor the deformation of ancient buildings within a set time period based on the registration results.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is loaded by the processor, it is able to perform the steps of the method according to any one of claims 1 to 6.

Images