树木三维点云的枝叶分割方法  

3D Point Cloud Segmentation Method for Branches and Leaves of Trees

作　　者：黄亮[1] 许文雅 谭帅 HUANG Liang;XU Wenya;TAN Shuai(Jiangsu Institute of Surveying and Mapping Engineering,Nanjing Jiangsu 210013,China;Jiangyin Greening Technology Supervision and Guidance Station,Jiangyin Jiangsu 214400,China;Jiangyin Landscape and Greening Management Institute,Jiangyin Jiangsu 214400,China)

机构地区：[1]江苏省测绘工程院,江苏南京210013 [2]江阴市绿化技术监督指导站,江苏江阴214400 [3]江阴市园林绿化管理所,江苏江阴214400

出　　处：《北京测绘》2022年第1期18-22,共5页Beijing Surveying and Mapping

基　　金：苏测院地理信息科研项目(SCYKY202006);江阴市城市园林绿化科研项目(JYYLKY202101)。

摘　　要：利用地面三维激光扫描点云构建树木模型时,对树木点云进行枝叶分割,可以提高单木分割、骨架提取、模型构建等步骤的准确性。针对现有枝叶分割方法参数设置复杂、需要训练样本、依赖颜色强度等不足,提出了一种新的基于邻域点分布规律的枝叶分割方法。该方法遍历树木点云中的激光点,首先利用K纬树算法搜索激光点的所有R邻域点,构成邻域点集;接着对邻域点集进行主成分分析,得到三对特征值和特征向量,利用特征值计算邻域点集的分布散乱度;然后将分布散乱度与预定义阈值比较,区分枝干和叶片的候选点;最后对枝干候选点实施欧氏聚类,利用聚类的点数和尺寸获取可靠的枝干点。使用江阴大桥公园的树木点云数据进行测试。结果表明:本文提出的树木点云枝叶分离方法准确率高、点云属性要求低、简单易用,可应用于多测合一绿化测量、城市绿地生物量统计、实景三维树木模型构建等领域。Using 3D point cloud to build tree model,the branch and leaf segmentation of tree point cloud can improve the accuracy of single tree segmentation,skeleton extraction,model construction and other steps.In view of the shortcomings of the existing branch and leaf segmentation methods,such as complex parameter setting,training samples and dependence on color or intensity,a new branch and leaf segmentation method based on the distribution of neighborhood points was proposed.The method traverses the laser points in the tree point cloud.Firstly,the k dimension tree algorithm was used to search all the R neighborhood points of the laser points to form the neighborhood point set;Next,the principal component analysis of the neighborhood point set was carried out to obtain three pairs of eigenvalues and eigenvectors,and the distribution scatter degree of the neighborhood point set was calculated by using the eigenvalues;Then the scatter degree was compared with the pre-defined threshold to distinguish the candidate points of branches and leaves.Finally,Euclidean clustering was applied to the selected branch points,and reliable branch points were obtained by using the number and size of cluster points.Using the tree point cloud data set of Bridge Park in Jiangyin City,the results showed that the method proposed in this paper had high accuracy,low requirements for point cloud attributes,and was simple and easy to use,which could be applied to the fields of greening measurement in multiple measurements integration,urban green space biomass statistics,and real 3D tree model construction.

关 键 词：三维点云 树木 枝叶分割 主成分分析 分布散乱度 

分 类 号：P258[天文地球—测绘科学与技术]