针对当前滤波算法在处理地形不连续区域或存在复杂建筑物区域时容易过分“腐蚀”地形并难以去除一些低矮植被的不足,提出了一种基于分割的机载LiDAR点云滤波算法。首先,对原始点云基于地表连续性进行分割;然后,在移除点数目较小的粗差点集之后采用对分割点集建立缓冲区的方法,区分地面和非地面点集;在较大地物经过迭代分割基本移除之后,使用约束平面的方法移除高度较小的地表附着物以实现滤波。实验结果表明,与经典滤波算法相比,该算法提高了地面点的分类精度,在滤除地物信息的同时能有效地保留地形特征。 In view of the fact that the current filtering algorithm is prone to over-eroded terrain and difficult to remove some low-lying vegetation when dealing with discontinuous areas or complicated building areas, a segmentation-based LiDAR point cloud filtering algorithm is proposed. Firstly, the original point cloud is segmented based on the continuity of the ground surface. Then, the method of establishing buffer zone for the set of segmentation points is adopted after removing the rough set of points with a small number of points to distinguish between ground and non-ground point sets. After the iterative segmentation is basically removed, a constrained plane approach is used to remove the smaller surface attachments for filtering. The experimental results show that compared with the classical filtering algorithm, the proposed algorithm improves the classification accuracy of ground points and effectively preserves the topographic features while filtering out the ground object information.