INNOVATIVE METHOD FOR POINT CLOUD FILTRATION, SEGMENTATION AND CLASSIFICATION TO DERIVE DIGITAL TERRAIN MODELS () BASED ON AIRBORNE LASER SCANNER () DATA

摘要

The invention relates to a method of point cloud data filtration, segmentation and classification for deriving digital terrain models based on the data obtained by airborne laser scanning technology. According to the invention, the method comprises a step of filtering the point clouds resulting from laser scanning, where an algorithm of hierarchical robust interpolation is used by creating an initial surface in which the points have equal individual weights, new values of the points weights being determined depending on the vertical point deviation from the most probable surface, a new surface is calculated by successive iterations and the points belonging to the terrain are identified, in order to accomplish a digital terrain model, followed by a step of segmentation of the point clouds during which the values of the normal lines in each point are calculated and the points are selected and analyzed depending on neighbourhoods, then a "region growing"-type segment forming process is applied by satisfying the similarity conditions, followed by extraction of the segments used in deriving the digital terrain model, and a step of classification of point clouds in which a comparative analysis of the terrain points obtained by robust interpolation and segmentation is made, the segments being classified as segments belonging to the terrain and segments not belonging to the terrain, to finally achieve an improved digital terrain model.