ABSTRACT
This chapter examines the extraction of the terrain surface from the Light Detection and Ranging (LiDAR) measurements, either the original point cloud or the already preprocessed one with terrain elevations sorted into a raster. The progressive densification and the surface-based filters use a surface in the generation of the classification. Progressive densification gradually adds points, building a digital terrain models (DTM) from a few points. The interpolated DTM can then be more accurate than a single point measurement. The performance of DTM data reduction is highly influenced by the existence of systematic and random measurement errors. The random measurement errors of the Airborne laser scanning points should rather be eliminated by applying a DTM interpolation strategy with measurement noise filtering capabilities. Stereoscopic viewing of the point cloud or the computed DTM can be an alternative giving more insight at the price of a longer procedure.
