Segmentation-based hierarchical interpolation filter using both geometric and radiometric features for LiDAR point clouds over complex scenarios

To enhance the filtering accuracy in complex environments, a segmentation-based hierarchical interpolation filter using both geometric and radiometric features is proposed in this paper. Specifically, raw point cloud is first segmented using DBSCAN with both geometric and radiometric features. Then, initial ground seeds are selected from the set of segments with the consideration of terrain features. Finally, all ground points are detected using an enhanced multiresolution hierarchical filter based on three reference ground surfaces of different attributes coupled with slope-adaptive thresholds. Four plots with complex landscapes were adopted to evaluate the results of the proposed method, and its accuracy was compared with those of seven state-of-the-art filtering methods. Results demonstrate that the proposed method obviously outperforms the classical filtering methods, with the reduction of average type I, II, and total errors by at least 15.1%, 10.0%, and 19.4%, respectively, and the improvement of the kappa coefficient by at least 2.9%.