车载LiDAR点云相连行道树精细分割  

Segmentation of connected street trees from mobile LiDAR data

作　　者：张西童 李永强[1] 毛杰[1] 李有鹏[1] 黄腾达[1] 

机构地区：[1]河南理工大学测绘与国土信息工程学院,河南焦作454000

出　　处：《测绘科学》2016年第8期111-115,共5页Science of Surveying and Mapping

基　　金：测绘地理信息公益性行业科研专项经费项目(201412020);国家十二五科技支撑项目(2012BAH34B00)

摘　　要：针对车载激光雷达获取的行道树树冠点云相连问题,该文提出了基于树木生长模型的距离加权分割方法。首先通过等距原则将相连树粗分割为单株树木,采用求差和拟合算法获得树木高度、胸径等属性信息;再根据树木的理论和实际生长模型,计算得到树冠距离加权值;最后在此基础上对树冠进行迭代处理,实现邻接行道树点云的精确分割。经过不同树种试验验证,该算法能够实现对多棵相连树点云的进一步精确分割。For the problem that point cloud of street tree obtained by mobile LiDAR is always connected,a distance-weighted segmentation method based on tree growth model was put forward in this paper.Firstly,trees that are connected together were segmented into single tree coarsely according to the principle of equidistance,then the attribute information of each tree,such as height,DBH,crown horizontal center location were obtained by differencing and fitting algorithms.Secondly,the distance weight values,which are the basis of point cloud segmentation between connected trees,were calculated based on tree theoretical and actual growth models.Finally,a finite iteration calculation was carried out by the repeating progress of crown diameter fitting,weight calculation and segmentation to realize the accurate segmentation of connected street trees.Experimental results verified the effectiveness of the proposed algorithm.

关 键 词：车载LiDAR 点云 行道树分割 生长模型 

分 类 号：P225.2[天文地球—大地测量学与测量工程]