Comparison of CSF and SMRF filtering methods for airborne LiDAR point cloud data

Abstract

Airborne LiDAR System (ALS) technologies are widely used for rapid data collection in a wide range of applications, including cultural heritage, Geography Information Systems (GIS), geodesy applications, 3D city modeling, and deformation analysis systems, and the generation of Digital Terrain Models (DTM). Filtering bare soil from point cloud data is critical for archaeologists, architects, and geomatics professionals employing airborne Light Detection and Ranging (LiDAR). Cloth Simulation Filtering (CSF) and Simple Morphological Filtering (SMRF), both ground filtering techniques, are discussed in this study. Airborne LiDAR point cloud data were split into the ground and non-ground point clouds for evaluation. A thorough evaluation of filtering accuracy necessitates comparing all point cloud data. However, because the data is so huge, this seems implausible. To adequately measure classification success, data manually identified as ground and non-ground was used as a reference. The performance of the CSF and SMRF approaches is enough, but it is impacted by point cloud type, slope, and vegetation type, according to our findings.