机载LiDAR点云中静态水体边界提取

Extraction of Static Water Boundary from Airborne LiDAR Point Cloud

从机载激光雷达点云中提取水体边界往往依赖影像辅助或假设水体表面点很少或回波强度很弱,而该假设与实际情况不一定相符。鉴于此,文章提出了一种仅基于点云坐标提取静态水体边界的方法。首先,构建Delaunay三角网,从三角网中提取空洞边界作为初始边界;其次,基于边界上点高程分布估计水面高程范围;然后,沿搜索方向进行迭代边界扩张,在扩张过程中动态更新水面高程估计,至边界不再变化;最后,修剪边界。利用江苏常州小黄山地区的数据进行了实验,100%提取到了主要水体,而且水体边界和正射影像中的边界套合良好。

The extraction of water boundaries from airborne LiDAR point clouds often relies on image assistance or the assumption that there are few water surface points or weak echoes,and this assumption does not necessarily match the actual situation.This paper proposes a method for extracting static water boundary based on point cloud coordinates only.Firstly,the Delaunay triangulation network is constructed,the hollow boundary is extracted from the network as the initial boundary.Secondly,the water surface elevation range is estimated based on the boundary point elevation distribution.And then,iterative boundary expansion is performed along the search direction.During the expansion process,the water surface elevation estimate is dynamically updated until the boundary no longer changes.Finally,the boundary is trimmed.Experiments are conducted using data from the Xiaohuangshan area in Changzhou,Jiangsu.100%of the main water bodies are extracted,and the obtained water body boundaries fit well with the boundaries in DOM.