Single Tree 3D Modeling From Terre Strial Laser Scanned Point Clouds

With the development of three dimensional laser scanner in ten years. People get the point cloud data of object has become increasingly simple,accurate. In the field of 3D building reconstruction,researching on 3D point cloud data has been quite extensive.In the 3D modeling of tree. By using the multi baseline digital close range photogrammetry system can obtain high accuracy of the trees the 3D point cloud model, but the terrestrial laser scanner can quickly and accurately to btain mass 3D point from tree surface. The point cloud data obtained by using the TLS can be used to shape and physiological parameters measurement and extraction of plant growth and used to construct the 3D model. The method of point cloud obtained by the TLS to build the 3D model of the trees, has become a hot esearch topic in the recent 10 years. It is not easy to reconstruct trees accurate 3D model from space scattered point cloud data. To obtain three-dimensional point cloud data of the surface needs to overcome a series of challenges by laser scanning:Mutual occlusion between the branches and leaves, inside the crown is not easy to be scanned and cause data loss. The breeze makes the outdoor living trees and leaves are swaying, To obtain the data error and noise.Because of the scanning characteristics and the morphological differences of trees, the point cloud data often contains noise and uneven distribution or error.These challenges have become the main research about constructing accurate 3D model based point cloud data of trees.In this paper, through the research of single tree 3D modeling echnology of live standing trees.We propose a simple method of the reconstruction in our project.The main research content is divided into two parts:(1)Three dimensional reconstruction of point cloud data of leaves.(2)Three dimensional reconstruction of point cloud data of the trunk.In the first part:First, from an entire tree point cloud data manually extracted from point cloud data of single leaves of different morphologies. Then carries on the curved surface fitting, denoising, edge extraction, triangulation operation. Finally completed the reconstruction of point cloud data of single leaf. Among them, this paper mainly researches on the link in the curved surface fitting. By using the method of least squares, orthogonal least squares method, moving least square method and three methods of surface fitting. Through the contrast experiment, The optimal method of curved surface fitting.In the second part:Study on manual cutting out of the trunk of point cloud data. The scattered point cloud data of the trunk surface fitting, denoising, skeleton extraction,3D modeling and other operations. Three dimensional reconstruction of the final completion of the trunk.