摘要
针对行道树点云复杂、中间夹杂较多杆状物等问题,提出一种基于体元邻域关系的行道树提取方法。以车载激光扫描点云数据构建的格网为单位进行地面点滤除与离散点降噪,然后对非空体元进行邻域分析实现树干点云提取,最后依据点云分层投影面积理论实现树冠点云提取。采用两组道路点云数据进行行道树点云提取试验,并将试验结果与人工提取行道树结果进行对比。结果表明,本文方法能自动化提取道路两侧行道树点云,行道树提取准确率达到90%以上,算法适应性较强。
In view of the complexity of the street trees point cloud and the inclusion of many rods in the middle,a method of extracting street trees based on the neighborhood relationship of voxel is proposed in this paper.Firstly,the ground point filtering and discrete point noise reduction are carried out in the unit of grid constructed by vehicle mounted laser scanning point cloud data.Secondly,the trunk point cloud is extracted by neighborhood analysis of non empty voxel,and the tree crown point cloud is extracted according to the theory of point cloud layered projection area.Finally,two groups of road point cloud data are used to carry out a street trees point cloud extraction test,and the test results are compared with the results of manually extracted street trees.The results show that the method in this paper can automatically extract the point cloud of street trees on both sides of the road,the accuracy of street trees extraction is more than 90%,and the algorithm has strong adaptability.
作者
裘嘉豪
郑奕宏
边奇海
QIU Jiahao;ZHENG Yihong;BIAN Qihai(Zhejiang Provincial Land Survey and planning Co.,Ltd.,Hangzhou,Zhejiang 310030,China)
出处
《测绘标准化》
2023年第1期50-54,共5页
Standardization of Surveying and Mapping
关键词
行道树提取
车载激光扫描
格网构建
体元
投影面积
street trees extraction
vehicle borne laser scanning
grid construction
voxel
projected area
