Vegetation filtering algorithm for UAV-borne lidar point clouds: a case study in the middle-lower Yangtze River riparian zone

A vegetation filtering algorithm is proposed for unmanned aerial vehicle UAV-borne lidar point clouds collected in the middle-lower Yangtze River riparian zone covered with multilayer and high-density vegetation. The proposed algorithm aims at generating digital elevation model, which consists of the following steps. First, multi-echo analysis is adopted for coarse filtering of point clouds. Then, morphological calculation is used to extract the ground seed points. Next, the trend surface of the local terrain is fitted. Meanwhile, the vegetation points are removed with the ground points preserved via random sample consensus. Experiments demonstrate that the performances of the algorithm were more accurate than Terrasolid’s TerraScan when dealing with UAV-borne lidar point clouds in limited sample of a multilayer and high-density vegetation covering areas. The research results also show that by using a special filtering approach it is possible to classify laser points into terrain and vegetation automatically even for thoroughly mixed vegetation and terrain points and penetration rate of below 15%.