Study On The Information Extraction Of Street Trees From Vehicle-borne LiDAR Point Cloud

As a rapid acquisition of spatial information technology,vehicle-borne LiDAR( Light Detecting and Ranging) has obvious advantages on high-precision of 3D information’s acquisition of objects in urban road environment and both sides of the street.Street tree is one of the most common object in road and city street scene, and also is an important part of the urban ecosystem.The distribution of trees scattered and shapes,and it’s a tedious and time-consuming work to get the information of trees manually.Extract trees’ information accurately and efficiently from vehicle-borne LiDAR is the goal of the researches’ attemption for a long time,it’s a difficult and hot issue to process the vehicle-borne LiDAR data.On the basis of in-depth understanding of vehicle-borne LiDAR system’s working principle and the objects’ expression in vehicle-borne LiDAR point cloud, the author take the city road scene for example,around the information extraction of trees in the scene launched a study.Study in this paper mainly includes the following three parts:The first part is point cloud’s classification and extraction of street trees.This part separate the street tree points from the original point cloud using the method of gradual separation, the sequence of the separation process can be respective:ground points filtered,objects clustering extraction of rod-like objects and separation of street tree point cloud.The second part is point cloud’s segmentation between trees and other connected rod-like objects.This part includes two parts:(1) ICP algorithm and reference sample were applied for effective removal of street light interspersed in the crown;(2) Aiming at the crown point cloud segmentation of the connected trees three methods were approached: the point cloud segmentation of crown based on distance, the point cloud segmentation of crown based on specific model and the point cloud segmentation of crown based on fitting crown,and do some tests and analysis respectively.The third part is the parameter information’s extraction of street trees. Based on the separation of single tree point cloud model realized the automatic extraction of street trees’ attribute information(center of tree’s bottom, tree height, trunk high, crown height, DBH and crown width);Based on the attribute information, studied the dynamic change of street trees, including the increase and growth of street trees.