A Self-Adaptive Morphological Filter without Consideration of Window Size for Airborne LiDAR Point Clouds

Abstract

Abstract Light detection and ranging (LiDAR) has the capability of rapidly collecting dense and accurate three-dimensional geospatial information, and therefore it is widely applied in various fields of geospatial applications. The morphological filtering approaches can filter non-ground points effectively, which is crucial for many tasks such as land cover classification and digital elevation model generation. A series of different windows are generally in need for removing non-ground objects with different sizes. In order to avoid the limitation of choosing the filtering windows, we adopt the geodesic transformations of mathematical morphology for filtering LiDAR point clouds. This algorithm enhances the robustness and automation without consideration of how to choose different windows. Experimental results demonstrate that this filtering algorithm is capable of effectively preserving terrain details and filtering various non-ground objects.