Three-dimensional tree geometric model reconstruction method based on trunk-like points

Abstract

The invention provides a three-dimensional tree geometric model reconstruction method based on trunk-like points. The three-dimensional tree geometric model reconstruction method specifically comprises the following steps: (1) segmenting a single tree; (2) constructing an initial tree skeleton; (3) optimizing the initial skeleton; (4) determining the branch radius; (5) adding tree leaves. The method has the advantages that (1) data loss of the point cloud is effectively processed; (2) the sensitivity of the algorithm to the point cloud density is reduced; and (3) the robustness of tree modeling is improved.

Description

technical field [0001] The invention relates to a method for reconstructing a three-dimensional tree geometric model based on similar trunk points, and belongs to the technical field of urban ground vegetation measurement and modeling. Background technique [0002] In the process of digitizing urban scenes, because trees, shrubs and other vegetation can be seen everywhere in urban scenes, fast and efficient reconstruction of tree 3D models is also an indispensable part of urban modeling. High-precision and realistic tree models It can make the digital urban scene more realistic and immersive. In addition, with the current development of big data and informatization, the process of digital forestry construction has been further accelerated. Building a more realistic tree model is also the basis and key to promote the process of forestry digitalization. [0003] LiDAR technology (Light Detection And Ranging, LiDAR) has the characteristics of large operation scene, fast spatia...

AI Technical Summary

Problems solved by technology

[0005] (1) Modeling based on the idea of ​​clustering can theoretically extract the skeleton points of any tree, but the time complexity of the clustering algorithm is high, and it changes exponentially with the increase of the number of points, which is not suitable for large data volume point clouds. Modeling research; in addition, clustering algorithms have a great dependence on the definition of point neighborhoods, and the construction of point neighborhood graphs directly affects the subsequent clustering process

[0006] (2) Modeling based on the graph theory method often has a great dependence on the resolution of the voxel, and the size of the voxel will have an impact on the calculation accuracy of the point cloud and the analysis of the neighborhood range: the voxel is too large, The accuracy of skeleton node extraction is reduced, and the voxel is too small, which will increase the calculation cost and weaken the calculation efficiency

[0007] (3) The constraint modeling method based on prior assumptions relies on the preprocessing process of denoising, and due to the induction of environmental factors such as natural wind and light in real scenes, the tree structure is not a completely regular cylinder or the cross section is not a regular ellipse body, the strict a priori assumption constraints have lost the accuracy of the tree model to a certain extent

[0008] (4) In the modeling strategy based on the Laplacian operator, the time complexity of the smoothing process of the operator is high, and the modeling efficiency is affected. When modeling with the Laplacian operator, the time complexity of the operator is often not considered degree and accuracy of skeleton line extraction

[0009] (5) The integrity and comprehensiveness of point cloud three-dimensional space information are weakened in the tree modeling strategy based on lightweight, without considering the skeleton, especially the internal branch skeleton of the crown, which will inevitably lead to inaccurate topological connections of branches and twig models. Loss, there are also partial topological link defects in fully expressing the information of trees, especially some details inside the canopy

Images