机载激光雷达点云的多层级地面滤波方法OA

After decades of development,existing ground filtering algorithms for point clouds can achieve satisfactory results in single specific scenes.However,these methods often lack robustness when processing point clouds in complex scenes,and each has its own advantages and limitations.To further improve filtering performance for point clouds in complex scenes,a multi-level ground filtering method for airborne light detection and ranging point clouds was proposed.First,noise removal was performed on the raw point cloud based on distance statistics.Second,a combination of cloth simulation and a morphologi-cal filtering algorithm was applied to rapidly extract a large number of ground seed points.Then,slope-adaptive interpolation was employed to supplement globally distributed ground seed points.Finally,an iterative densification process based on trian-gular irregular networks was used to extract the remaining ground points.To validate the effectiveness of the proposed method,point clouds in four complex scenes were tested.The results demonstrate that the proposed approach significantly improves robustness and accuracy compared to a single algorithm,achieving an average total error rate in average filtering of less than 3%.The method exhibits strong adaptability in complex scenes containing steep slopes,dense vegetation,and large buildings.