摘要
本文提出了一种从城区车载激光雷达点云数据中提取行道树的方法。该方法先为每个扫描点计算局部几何特征，将扫描点粗标记为线状点、面状点、球状点；然后根据扫描点的标记进行欧氏聚类分割；最后通过计算每个分割块的空间特征实现行道树的精确提取。试验结果表明，该方法不仅能够提取较为完整的树木，而且对因遮挡导致的扫描数据不完整的树木亦能正确提取，具有较高的提取正确率和较好的完整性。
In this paper, a new method for extracting street trees from MLS point cloud data in urban area is proposed. Firstly, scanning points are marked as linear, planar or spherical points based on the calculation results of local geometrical features. Secondly, Euclidean clustering segmentation is conducted according to the markings of scanning points. Finally, the accurate extraction of street trees is realized by calculating the spatial features of each segment. The test results show that the method can extract trees with and without complete scanning data due to blocking with better accuracy and integrity.
作者
姚强强
董广军
邵磊
于英
Yao Qiangqiang;Dong Guangjun;Shao Lei;Yu Ying(Information Engineering University, Zhengzhou 450001, China)
出处
《测绘科学与工程》
2018年第1期49-54,共6页
Geomatics Science and Engineering
基金
国家自然科学基金资助项目（41501482,61272146）.
关键词
车载激光点云
局部几何特征
欧氏聚类分割
行道树提取
MLS point cloud
local geometrical feature
Euclidean clustering segmentation
street tree extraction
