摘要
提出了一种建筑物LiDAR点云的屋顶边界提取方法。首先构建了离散的建筑物屋顶LiDAR点的TIN模型,在TIN模型中根据点的空间几何关系,过滤整个三角网的边界线,从而过滤出初始边界点。在初始边界点构成的边界中,过滤出边界斜率变化显著的点作为拐点。利用所有的屋顶LiDAR点将拐点扩展,得到边界的扩展点,由扩展点构成的屋顶边界为最终提取的建筑物屋顶边界。
This paper presents a method for boundary extraction of building roof from LiDAR point cloud. Firstly, a TIN model of discrete building roof LiDAR point cloud is built. Based on spatial geometry relationship in the TIN model, after some triangles with too larger side lengths than the average point distance are deleted, the borderlines of residual irregular triangulated network can be extracted to obtain initial boundary points. From the initial boundary points, points which two vectors have distinct change of slope are filtered as inflexion points. Some LiDAR points might locate outside the polygon constructed by inflexion points. So the inflexion points need to be expanded using all building roof LiDAR points. The final building roof boundary is constructed by the expanded points. The experimental result shows that this method is able to effectively extract all kinds of polygon building roof boundary.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2009年第4期383-386,共4页
Geomatics and Information Science of Wuhan University
基金
上海市教委重点学科建设资助项目(J50104)
上海市教委科研发展基金资助项目(07ZZ09)
