3D Point Cloud Of The Feature Extracation And Tree Reconstruction

Accurate branch-dimensional reconstruction of plants is one of the key research issues in computer graphics in recent years,and as three-dimensional scanning equipment has become more and more low-cost and high-precision,people have gradually applied threedimensional reconstruction in various fields,such as digital city,preservation and restoration of cultural relics,and real scene simulation in virtual reality.Many scholars have also conducted in-depth research on the reconstruction of branches,they put forward various methods,and achieved certain results.However,due to the various shapes of branchs,current reconstruction methods cannot fully meet the requirements for obtaining three-dimensional feature information of branchs characteristics.Therefore,the method proposed in this paper through the geometric reconstruction of three-dimensional point cloud trees provides a reference basis for plant digital information for other fields such as digital city.In this paper,the branch model is used as the experimental object,the structured light system is used to obtain the surface point cloud information of the branch model.We propose a method of branch point cloud feature extraction and reconstruction based on the cylindrical coordinate system.In this paper,we mainly research on point cloud preprocessing,point cloud registration,point cloud segmentation,hole repair and model reconstruction.Point cloud preprocessing includes point cloud denoising and degeneration;Firstly,we use the ICP algorithm for fine registration of point clouds to ensure that the complete brachdimensional information of the branch point cloud is obtained;Secondly,we establish the three-dimensional point cloud topological relationship by using the rectangular coordinate system to convert the cylindrical coordinate system,and then extract features of the branch point cloud and estimate the branch point position based on the cylindrical coordinate system;Thirdly,and using the branch cross-sectional feature to determine the branch branch point,combining location of the branch point with the Euclidean distance segmentation algorithm to achieve the segmentation of the branch model;Finally,the B-spline surface reconstruction is performed on the point cloud that has been filled with holes.The method proposed in the article is based on cylindrical coordinates,which is more accurate for the feature extraction and segmentation of point cloud branches.This method can be widely used in various target models with cylindrical geometric characteristics.