Multidirectional Shift Rasterization (MDSR) Algorithm for Effective Identification of Ground in Dense Point Clouds

With the ever-increasing popularity of unmanned aerial vehicles and other platforms providing dense point clouds, filters for identification ground points in such clouds are needed. Many have been proposed widely used, usually based on determination an original surface approximation subsequent within a predefined distance from surface. We presented new filter, multidirectional shift rasterization (MDSR) algorithm, which is different principle, i.e., just lowest individual grid cells, shifting along both planar axis tilting entire grid. The principle was detail visually numerically compared with commonly used (PMF, SMRF, CSF, ATIN) three sites ruggedness vegetation density. Visually, MDSR filter showed smoothest thinnest profiles, ATIN only comparably performing. same confirmed when comparing filtered by MDSR-based goodness fit cloud demonstrated root mean square deviations (RMSDs) found below MDSR-generated (ranging, depending site, between 0.6 2.5 cm). In conclusion, this paper introduced newly developed that outstandingly performed at all sites, identifying great accuracy while filtering out maximum above-ground outperforming aforementioned filters. dilutes somewhat; however, can be perceived as benefit rather than disadvantage.