A vectorized skeleton extraction method and system for outdoor scene point cloud data

Abstract

The present invention discloses a method and system for extracting vectorized skeletons from outdoor scene point cloud data. The method comprises the following steps: S01. acquiring point cloud data and panoramic image data of a scene captured in an outdoor scene; S02. extracting edge points from the point cloud data and the panoramic image data respectively; S03. constructing a sparse graph structure using a point cloud edge point set to obtain a point cloud sparse skeleton, and constructing a dense graph structure using an image edge point set to obtain an image dense vectorized skeleton; S04. spatially transforming the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton; S05. fusing the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted. The present invention does not rely on prior information about geometric shapes and can quickly and accurately extract the vectorized skeleton of point cloud data.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a method and system for extracting vectorized skeletons from outdoor scene point cloud data. Background Art

[0002] A point cloud is a dataset in which each point represents a set of X, Y, and Z geometric coordinates and an intensity value, which records the strength of the return signal based on the object's surface reflectivity. When these points are combined, they form a point cloud, a collection of data points representing a 3D shape or object in space. Point clouds can be automatically colored for more realistic visualization.

[0003] In practical applications, the amount of point cloud data obtained is usually large, and point cloud simplification is required. In the point cloud simplification task, it is necessary to extract the skeleton structure of the dense point cloud as a lightweight intermediate modality for understanding and rendering. To achieve point cloud skeleton extraction, the following two methods are generally used in existing technologies:

[0004] 1) Semantic-based point cloud skeletonization method

[0005] This type of method performs upsampling on the point cloud, that is, taking a point cloud as input and outputting a denser point cloud that falls on the geometry (such as a surface) implied by the input point cloud. Semantic-based point cloud skeletonization methods learn features for each point at multiple granularities (from local to global), then reduce the point set in the feature space, and finally map the reduced point set back into three dimensions to achieve point cloud skeletonization.

[0006] 2) Geometry-based point cloud skeletonization method

[0007] This type of method extracts the edge features of the points or the knn method, aggregates points with similar features or located in the same area into one point, and forms a sparse point cloud skeleton.

[0008] However, whether it is a semantic-based point cloud skeletonization method or a geometric-based point cloud skeletonization method, the above-mentioned point cloud skeleton extraction methods in the prior art have the following problems:

[0009] 1. Both methods rely on a complex object modeling process and have high requirements for the geometric shape of the model itself. However, in actual working conditions, it is difficult to obtain complete and accurate semantic prior information for the initial point cloud obtained by scanning in a short time and limited environment. As a result, the point cloud densification modeling is not accurate and lacks robustness. Therefore, the actual point cloud skeleton extraction accuracy is not high. In addition, the computational complexity of the above point cloud skeleton extraction methods is high and requires a lot of computing resources.

[0010] 2. Since the outdoor environment is an open environment, it is easily affected by environmental interference. The deviation of depth images collected in outdoor environments is large, and it is impossible to use depth images to assist in skeleton extraction. Summary of the Invention

[0011] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a vectorized skeleton extraction method and system for outdoor scene point cloud data, which does not rely on prior information of geometric shapes, has a simple implementation method, small computational complexity, high extraction accuracy and efficiency, and strong robustness. The method and system can be used to realize fast and accurate vectorized skeleton extraction of scene point cloud data in non-closed outdoor environments.

[0012] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0013] A vectorized skeleton extraction method for outdoor scene point cloud data, comprising the following steps:

[0014] Step S01. Acquire point cloud data and panoramic image data of a scene captured in an outdoor scene;

[0015] Step S02: extracting edge points from the point cloud data and the panoramic image data to obtain a point cloud edge point set and an image edge point set;

[0016] Step S03. Using the point cloud edge point set to construct a sparse graph structure to obtain a point cloud sparse skeleton, and using the image edge point set to construct a dense graph structure to obtain an image dense vectorized skeleton;

[0017] Step S04: performing spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton;

[0018] Step S05: Fusing the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted.

[0019] Furthermore, when edge points are extracted from the point cloud data in step S02, the distance between each point in the point cloud data and the remaining points is calculated. When the minimum distance between the target point and the remaining points is greater than a first preset threshold, , then the marked target point is an edge point, and the edge points obtained by all the marks constitute the point cloud edge point set.

[0020] Furthermore, when edge points are extracted from the panoramic image data in step S02, the distance between the color value of each pixel in the panoramic image data and the surrounding pixels is calculated. When the minimum value of the distance between the color value of the target pixel and the surrounding pixels is greater than the second preset threshold When , the target pixel point is marked as an edge point, and all the marked edge points constitute the image edge point set.

[0021] Furthermore, in step S03, the sparse graph structure is constructed using the point cloud edge point set to obtain a point cloud sparse skeleton, which includes: constructing a first graph structure ,and ,in Represents a vertex, each vertex is the three-dimensional space coordinate of each edge point in the point cloud The vectorized representation of The value of each edge is the value of the corresponding edge point in the entire point cloud data. The minimum Euclidean distance between the points in and the rest of the points is constructed by the first graph structure The point cloud sparse skeleton is obtained.

[0022] Furthermore, in step S03, the dense graph structure is constructed using the image edge point set to obtain the dense vectorized skeleton of the image, which includes: constructing a second graph structure ,and ,in Represents vertices, each vertex is selected from the original image data The position coordinates of 25 adjacent points in The vectorized representation of The value of each edge is the color value of the 25 adjacent points of the corresponding edge point in the original image data. The average value of the second graph structure constructed by A dense vectorized skeleton of the image is obtained.

[0023] Furthermore, in step S04, the point cloud sparse skeleton and the image dense vectorized skeleton are spatially transformed by using the transformation parameters rotation matrix R and translation vector t between the point cloud data and the panoramic image data, wherein the point cloud sparse skeleton is spatially transformed according to Get the sparse skeleton of the transformed point cloud , Represents the sparse skeleton of the point cloud before transformation, and the dense vectorized skeleton of the image is transformed according to Get the dense vectorized skeleton of the transformed image , Represents the dense vectorized skeleton of the image before transformation.

[0024] Furthermore, the step S05 includes: constructing a complete graph structure G, in which the vertices are composed of a combination of vertices of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the edges are composed of a combination of edges of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the complete point cloud vectorized skeleton to be extracted is obtained from the complete graph structure G.

[0025] Furthermore, in the complete graph structure G, the vertex V is specifically get, are the vertices of the transformed point cloud sparse skeleton, The vertices and edges of the dense vectorized skeleton of the transformed image Specific according to get, The edges of the sparse skeleton of the transformed point cloud, The edges of the densely vectorized skeleton of the transformed image.

[0026] A vectorized skeleton extraction system for outdoor scene point cloud data, comprising:

[0027] A data acquisition module is used to acquire point cloud data and panoramic image data of the scene in an outdoor scene;

[0028] An edge point extraction module is used to extract edge points from the point cloud data and the panoramic image data respectively to obtain a point cloud edge point set and an image edge point set;

[0029] A skeleton extraction module is configured to construct a sparse graph structure using the point cloud edge point set to obtain a point cloud sparse skeleton, and to construct a dense graph structure using the image edge point set to obtain an image dense vectorized skeleton;

[0030] A spatial transformation module is used to perform spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton;

[0031] The skeleton fusion module is used to fuse the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted.

[0032] A vectorized skeleton extraction system for outdoor scene point cloud data includes a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0033] Compared with the existing technology, the advantages of the present invention are: the present invention can improve the authenticity and comprehensibility of the final output by separately acquiring point cloud data and panoramic image data for fusion processing, and at the same time, combine the graph structure to construct sparse and dense skeletons of point cloud and panoramic image data respectively. Compared with the traditional method of directly aggregating similar semantic points, it can effectively retain spatial information, thereby ensuring the accuracy of skeleton extraction. The use of a graph structure in the form of vectorized representation of point cloud data can also effectively reduce the amount of calculation of the method and improve the robustness of point cloud skeleton extraction. There is no need to rely on prior information such as the geometric shape of the model, that is, it can realize fast and accurate skeleton extraction of high-density three-dimensional point clouds in outdoor scenes based on images. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 3 is a schematic diagram of the implementation process of the vectorized skeleton extraction method for outdoor scene point cloud data in this embodiment.

[0035] Figure 2 3 is a structural diagram of the vectorized skeleton extraction system for outdoor scene point cloud data in this embodiment. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the scope of protection of the present invention is not limited thereby.

[0037] Taking into account that point cloud data can provide three-dimensional spatial characteristics and panoramic image data has spatial directionality and optical characteristics, the fusion of point cloud data and panoramic image data can obtain more complete and accurate information. Considering the spatial characteristics of point cloud data in outdoor scenes, the graph structure has spatial structure and can retain spatial information. The present invention aims at the problem of point cloud skeleton extraction in outdoor scenes. By separately obtaining point cloud data and panoramic image data for fusion processing, the authenticity and comprehensibility of the final output can be improved. At the same time, sparse and dense skeletons of point cloud and panoramic image data are constructed respectively in combination with the graph structure. Compared with the traditional method of directly aggregating similar semantic points, it can effectively retain spatial information, thereby ensuring the accuracy of skeleton extraction. The use of a graph structure in the form of vectorized representation of point cloud data can also effectively reduce the amount of computation of the method and improve the robustness of point cloud skeleton extraction. There is no need to rely on prior information such as the geometric shape of the model, that is, fast and accurate skeleton extraction of high-density three-dimensional point clouds can be achieved based on images.

[0038] like Figure 1 As shown, the steps of the vectorized skeleton extraction method for outdoor scene point cloud data in this embodiment include:

[0039] Step S01. Acquire point cloud data and panoramic image data of a scene captured in an outdoor scene;

[0040] Step S02: extract edge points from the point cloud data and the panoramic image data to obtain a point cloud edge point set and an image edge point set;

[0041] Step S03. Using the point cloud edge point set to construct a sparse graph structure to obtain a point cloud sparse skeleton, and using the image edge point set to construct a dense graph structure to obtain an image dense vectorized skeleton;

[0042] Step S04: performing spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton;

[0043] Step S05: Fusing the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted.

[0044] In this embodiment, the outdoor scene is an open, non-enclosed scene. Point cloud data and panoramic image data of a target are collected separately at the same location in the outdoor scene. The point cloud data can be used to obtain the target's three-dimensional spatial characteristics, while the panoramic image data can be used to obtain the target's spatial directionality and optical characteristics. Fusion processing of the point cloud data and panoramic image data can then be performed to obtain more complete and accurate information, improving the authenticity and comprehensibility of the final output. The method for collecting the point cloud data and panoramic image data in step S01 is not limited and can be selected based on actual needs.

[0045] In this embodiment, when edge points are extracted from the point cloud data in step S02, the distance between each point in the point cloud data and the remaining points is calculated. When the minimum distance between the target point and the remaining points is greater than a first preset threshold, , then the marked target point is the edge point, and the edge points obtained by all the marks constitute the point cloud edge point set.

[0046] Assuming that the scene point cloud data collected is P and the panoramic image data is I, the size of the collected data is Point cloud data To extract edge points, the specific extraction method is: for each point Coordinates, calculate the Euclidean distance between each point and the remaining N-1 points, when the minimum distance between a point and the remaining N-1 points is greater than the first preset threshold When , this point is marked as an edge point, and the point cloud composed of all edge points is recorded as , which is the point cloud edge point set, the specific size is .

[0047] In this embodiment, when edge points are extracted from the panoramic image data in step S02, the distance between the color value of each pixel in the panoramic image data and the surrounding pixels is calculated. When the minimum value of the distance between the color value of the target pixel and the surrounding pixels is greater than the second preset threshold, When , the target pixel is marked as an edge point, and all the marked edge points constitute the image edge point set.

[0048] Take the collected panoramic image with the size of H×W I Taking edge point extraction as an example, the specific extraction method is: the color value of each pixel The Euclidean distance is calculated with the surrounding 24 pixels. When the minimum distance between a pixel and the surrounding 24 pixels is greater than the second preset threshold, When , mark this point as an edge point, and record the image point set composed of all edge pixels as , which is the image edge point set, the specific size is .

[0049] In this embodiment, step S03 uses the edge point set of the point cloud to construct a sparse graph structure to obtain a sparse skeleton of the point cloud, which includes: (Specific size is ) Build the first graph structure ,and ,in Represents a vertex, each vertex is the three-dimensional space coordinate of each edge point in the point cloud The vectorized representation of The value of each edge is the value of the corresponding edge point in the entire point cloud data. The minimum Euclidean distance between the points and the rest of the points, thus constructing a sparse graph structure , the first graph structure constructed by That is, the sparse skeleton of the point cloud is obtained.

[0050] In this embodiment, step S03 uses the image edge point set to construct a dense graph structure to obtain the image dense vectorized skeleton, which includes: (Specific size is ) Build the second graph structure , ,in represents a vertex, For the edge, for the image point set For each pixel point in the original image data, each vertex selects its The position coordinates of 25 adjacent points in The vector representation of each edge is the color value of the 25 adjacent points of the corresponding edge point in the original image data. The average value of , in this way a dense graph structure is constructed , the second graph structure constructed by That is, the dense vectorized skeleton of the image is obtained.

[0051] Point cloud data is discrete data with a small amount of data, while panoramic data is continuous data with a large amount of data. The pixel density of panoramic data is much greater than the point density of point cloud. This embodiment constructs a sparse graph structure for point cloud data and a dense graph structure for panoramic data. It can fully combine the density characteristics of point cloud and panoramic data to construct an effective graph structure, thereby utilizing the spatial structure of the graph structure to accurately construct a skeleton for point cloud semantics.

[0052] In this embodiment, in step S04, the point cloud sparse skeleton and the image dense vectorized skeleton are spatially transformed by using the transformation parameters rotation matrix R and translation vector t between the point cloud data and the panoramic image data, wherein the point cloud sparse skeleton is specifically transformed according to Get the sparse skeleton of the transformed point cloud , Represents the sparse skeleton of the point cloud before transformation, and the dense vectorized skeleton of the image is specifically Get the dense vectorized skeleton of the transformed image , The transformation parameters between the point cloud and the image, the rotation matrix R and the translation vector t, can be set arbitrarily according to the actual needs such as the degree of transformation required.

[0053] In this embodiment, step S05 specifically includes: constructing a complete graph structure G, wherein the vertices are composed of the vertices of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the edges are composed of the edges of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the complete point cloud vectorized skeleton to be extracted is obtained from the complete graph structure G. In the above complete graph structure G, the vertex V is specifically formed according to get, are the vertices of the transformed point cloud sparse skeleton, The vertices and edges of the dense vectorized skeleton of the transformed image Specific according to get, The edges of the sparse skeleton of the transformed point cloud, is the edge of the dense vectorized skeleton of the transformed image. and The fusion is performed to construct a complete graph structure G, from which the skeleton structure of the original point cloud P is obtained, which can be used for subsequent operations such as rendering and processing. Because the complete graph structure G integrates both point cloud data and panoramic image data, and because the spatial structure characteristics of the graph structure in the complete graph structure G preserve the spatial information of both point cloud data and panoramic image data, the point cloud skeleton can be extracted quickly and accurately.

[0054] like Figure 2 As shown, the vectorized skeleton extraction system for outdoor scene point cloud data in this embodiment includes:

[0055] A data acquisition module is used to acquire point cloud data and panoramic image data of the scene in an outdoor scene;

[0056] The edge point extraction module is used to extract edge points from the point cloud data and the panoramic image data respectively to obtain the point cloud edge point set and the image edge point set;

[0057] A skeleton extraction module is used to construct a sparse graph structure using the edge point set of the point cloud to obtain a sparse skeleton of the point cloud, and to construct a dense graph structure using the edge point set of the image to obtain a dense vectorized skeleton of the image;

[0058] A spatial transformation module is used to perform spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton;

[0059] The skeleton fusion module is used to fuse the sparse skeleton of the transformed point cloud and the dense vectorized skeleton of the transformed image to obtain the complete point cloud vectorized skeleton to be extracted.

[0060] In this embodiment, the edge point extraction module includes a point cloud edge point extraction unit for extracting edge points from point cloud data, and an image edge point extraction unit for extracting edge points from panoramic image data. The skeleton extraction module specifically includes a point cloud skeleton extraction unit for constructing a sparse graph structure using a point cloud edge point set to obtain a sparse point cloud skeleton, and an image skeleton extraction unit for constructing a dense graph structure using an image edge point set to obtain a dense vectorized image skeleton.

[0061] The vectorized skeleton extraction system for outdoor scene point cloud data in this embodiment corresponds one-to-one to the above-mentioned vectorized skeleton extraction method for outdoor scene point cloud data, and will not be described in detail here.

[0062] In another embodiment, the vectorized skeleton extraction system for outdoor scene point cloud data of the present invention may also include a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0063] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions described in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which implements the functions specified in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0064] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A vectorized skeleton extraction method for outdoor scene point cloud data, characterized in that the steps include: Step S01. Acquire point cloud data and panoramic image data of a scene captured in an outdoor scene; Step S02: extracting edge points from the point cloud data and the panoramic image data to obtain a point cloud edge point set and an image edge point set; Step S03. Using the point cloud edge point set to construct a sparse graph structure to obtain a point cloud sparse skeleton, and using the image edge point set to construct a dense graph structure to obtain an image dense vectorized skeleton; Step S04: performing spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton; Step S05: Fusing the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted.

2. The vectorized skeleton extraction method for outdoor scene point cloud data according to claim 1, characterized in that: When edge points are extracted from the point cloud data in step S02, the distance between each point in the point cloud data and the remaining points is calculated. When the minimum distance between the target point and the remaining points is greater than a first preset threshold, , then the marked target point is an edge point, and the edge points obtained by all the marks constitute the point cloud edge point set.

3. The vectorized skeleton extraction method for outdoor scene point cloud data according to claim 1, characterized in that: When edge points are extracted from the panoramic image data in step S02, the distance between the color value of each pixel in the panoramic image data and the surrounding pixels is calculated. When the minimum value of the distance between the color value of the target pixel and the surrounding pixels is greater than the second preset threshold When , the target pixel point is marked as an edge point, and all the marked edge points constitute the image edge point set.

4. The vectorized skeleton extraction method for outdoor scene point cloud data according to claim 1, characterized in that: The step S03 uses the point cloud edge point set to construct a sparse graph structure to obtain a point cloud sparse skeleton, which includes: constructing a first graph structure ,and ,in Represents a vertex, each vertex is the three-dimensional space coordinate of each edge point in the point cloud The vectorized representation of The value of each edge is the value of the corresponding edge point in the entire point cloud data. The minimum Euclidean distance between the points in and the rest of the points is constructed by the first graph structure The point cloud sparse skeleton is obtained.

5. The vectorized skeleton extraction method for outdoor scene point cloud data according to claim 1, characterized in that: In step S03, the dense graph structure is constructed using the image edge point set to obtain a dense vectorized image skeleton, which includes: constructing a second graph structure ,and ,in Represents a vertex, each vertex selects each edge point in the original image data The position coordinates of 25 adjacent points in The vectorized representation of The value of each edge is the color value of the 25 adjacent points of the corresponding edge point in the original image data. The average value of the second graph structure constructed by A dense vectorized skeleton of the image is obtained.

6. The vectorized skeleton extraction method for outdoor scene point cloud data according to any one of claims 1 to 5, characterized in that: In the step S04, the point cloud sparse skeleton and the image dense vectorized skeleton are spatially transformed by using the transformation parameters between the point cloud data and the panoramic image data, the rotation matrix R and the translation vector t, wherein the point cloud sparse skeleton is spatially transformed according to Get the sparse skeleton of the transformed point cloud , Represents the sparse skeleton of the point cloud before transformation, and the dense vectorized skeleton of the image is transformed according to Get the dense vectorized skeleton of the transformed image , Represents the dense vectorized skeleton of the image before transformation.

7. The vectorized skeleton extraction method for outdoor scene point cloud data according to any one of claims 1 to 5, characterized in that: The step S05 includes: constructing a complete graph structure G, wherein the vertices are composed of a combination of vertices of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the edges are composed of a combination of edges of the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton, and the complete point cloud vectorized skeleton to be extracted is obtained from the complete graph structure G.

8. A vectorized skeleton extraction system for outdoor scene point cloud data, characterized in that: include: A data acquisition module is used to acquire point cloud data and panoramic image data of the scene in an outdoor scene; An edge point extraction module is used to extract edge points from the point cloud data and the panoramic image data respectively to obtain a point cloud edge point set and an image edge point set; A skeleton extraction module is configured to construct a sparse graph structure using the point cloud edge point set to obtain a point cloud sparse skeleton, and to construct a dense graph structure using the image edge point set to obtain an image dense vectorized skeleton; A spatial transformation module is used to perform spatial transformation on the point cloud sparse skeleton and the image dense vectorized skeleton to obtain a transformed point cloud sparse skeleton and a transformed image dense vectorized skeleton; The skeleton fusion module is used to fuse the transformed point cloud sparse skeleton and the transformed image dense vectorized skeleton to obtain the complete point cloud vectorized skeleton to be extracted.

9. A vectorized skeleton extraction system for outdoor scene point cloud data, comprising a processor and a memory, wherein the memory is used to store a computer program, characterized in that: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 7.

Images