密集低矮植被区LiDAR点云地面滤波算法

LiDAR point cloud ground filtering algorithm in dense and low vegetation area

针对地形复杂且低矮植被茂密的矿区LiDAR点云特点,本文提出了一种基于坡度信息并结合平面拟合的地面滤波算法。该方法采用二级格网法逐级选取地面种子点,在每个一级格网中,利用地面种子点通过最小二乘拟合法进行平面拟合并构建地面模型,最后达到区分地面点和非地面点的效果。与传统坡度法和布料模拟法的对比试验表明,该方法能够有效滤除密集低矮灌木,以及较好地保留较大坡度地形。

Aiming at the characteristics of LiDAR point cloud in mining areas with complex terrain and dense low vegetation,this paper proposes a ground filtering algorithm based on slope information combined with plane fitting.This method uses the second-level grid method to select ground seed points by hierarchical extraction,and uses the ground seed points in each first-level grid to construct a ground model by plane fitting through the least squares fitting method.Finally,ground point and non-ground point can be distinguished.Through comparative experiments with the traditional slope filter and cloth simulation filter,it is concluded that this method can effectively filter out dense and low shrubs and can better retain the larger slope terrain.