基于机载LiDAR数据的建筑物屋顶边界提取

Building roof boundary extraction based on airborne LiDAR data

针对机载Li DAR点云数据提出了一种建筑物屋顶边界提取方法。首先,将建筑物屋顶激光点云投影至XOY平面上,对屋顶点云构建网孔,初步提取建筑物边缘点云并去除部分孤立点;然后,采用改进的带距离控制的卷包裹算法对角点进行提取,根据相邻房屋边缘点一侧不存在激光点的性质,对边缘点云进行二次提取;最后,对房屋边线上的点云采用最小二乘直线拟合方法分别拟合屋顶各边线,并对拟合后的边线进行规则化及扩展,还原建筑物边界线。相比于基于不规则三角网模型的边缘提取方法,采用本文算法获得了更多边缘点,并且避免了凸包性缺陷导致对建筑物边缘提取造成的关键边缘点遗漏问题,对各种形状建筑物均有较强适应性,且运行效率更高。

From airborne LiDAR point cloud data,this paper proposes a building roof boundary extraction method.First,building roof laser point cloud is projected onto the XOY plane.Mesh holes are built on the roof point cloud,and the building edge point cloud is extracted to remove some outliers.According to the principle of roll-wrap algorithm with distance limit,the corner point of the extraction is performed.The edge point clouds are secondary extracted from the neighboring corners according to the nature of the absence of laser spots on the edge of the neighboring house.Finally,the least squares line is used for the point clouds on the sidelines of the house.The fitting method fits each sideline of the house separately,regularizes and expands the fitted edge,and restores the building boundary line.Finally,it extracts more edge points than the TIN model,avoiding the impact of convexity on the extraction of building edges.The algorithm is more adaptable than other kinds of shape buildings with higher efficiency.