面向地面点识别的机载LiDAR点云分割方法研究

An airborne LiDAR point cloud segmentation method for recognizing the ground measurments

提出一种面向地面目标识别的机载LiDAR点云分割方法。方法首先求每个激光脚点的法向量和残差,由此确定种子点和种子平面;然后对种子点进行区域生长,生长的过程中以邻接点到种子平面的距离和邻接点与种子点的法向量角度差作为相似性的度量标准;当全部的扫描点都被划分,则算法终止。实验表明,文中提出的分割方法,对于城区区域和农村区域的地面目标有很好的识别效果。

A region-growing-based airborne LiDAR point cloud segmentation method is proposed to extract the ground measurements in the point clouds.Particularly,the normal and residual for each point is estimated by fitting aplane to some neighboring points,and the seed points and seed planes are determined by the above two features.The region growing process is performed from the seed points,where the distance between the neighbors to the current seed plane and the angle difference between the normal of the current seed and its neighbors are the two criterion for judging the similarity.The experiments show that the proposed method is capable of better recognizing the ground measurements for both the urban regions and the natural regions.