摘要
针对车载激光雷达点云初始聚类中心难以确定的问题,该文提出了一种基于最大网格密度的近邻聚类算法对点云实现分割,并以高程、法向量和投影密度作为约束条件对分割后的点云块进行地物的分类识别。通过对车载激光雷达的部分点云数据进行相关试验,结果表明该方法可以精确有效地实现城市典型地物分类。
Aiming at the problem that initial clustering center is hard to determine, this paper put for- ward a segmentation algorithm of point cloud based on maximum grid density. Then the elevation and vec- tors were used as constraints to achieve the classification of different features. Experimental result showed that this method would be accurate and effective for the classification of typical city features.
出处
《测绘科学》
CSCD
北大核心
2016年第4期77-82,共6页
Science of Surveying and Mapping
基金
山东省自然科学基金项目(ZR2014DM014)
十二五国家科技支撑计划课题项目(2012BAH27B04)
海岛(礁)重点实验室开放基金项目(2013B09)
测绘学院科研创新团队支持计划项目
关键词
车载LiDAR
聚类
点云分割
分类
vehicle-borne LiDAR
cluster
point cloud segmentation
classification
