Extraction of ground points from LiDAR data based on slope and progressive window thresholding (SPWT)

Filtering of airborne LiDAR point clouds has broad applications, such as Digital Terrain Model (DTM)generation and threedimensional urban modeling. Although several methods have been developed toseparate the point clouds into ground and nonground points, there are some challenges to identify thecomplex objects such as bridge and eccentric roofs. In this study, a new algorithm based on the Slope andProgressive Window Thresholding (SPWT) is proposed for ground filtering of LiDAR data. This algorithmis based on both multiscale and slope methods that have strong effects on filtering the LiDAR data. Theproposed algorithm utilizes the slope between adjacent points and the elevation information of points in alocal window to detect nonground objects. Therefore, not only it benefits from vertical information in eachlocal window to detect the nonground points, but it also uses the neighbor information in directionalscanning, and it prevents the errors introduced by the sensitivity to direction. According to the physicalcharacteristics of the ground surface and the size of objects, the best threshold values are considered. Inorder to evaluate the performance of the SPWT method, both low and high resolution datasets were appliedthat their average overall accuracy were reported to be 94.21% and 93.08%, respectively. These resultsproved that, irrespective of data resolution, the SPWT method could effectively remove the nongroundpoints from airborne LiDAR data.