Extracting 3D contour features of urban scenes from ground-based laser range data

3D features of urban scenes are the most important basic elements, we present a method for extracting 3D features of urban scenes from ground-based laser range data. We give depth image representation, define vertical and horizontal scanline, and remove the invalid data including trees, cables, fences, cars, passers-by, outlies and so on. We employ local self-adaptive curve fitting to extract line segments from main architecture structures. We introduce virtual feature conception for the first time, which indicates some implying structure features of urban scenes. The approach described in this paper has been implemented and applied to create the reconstruction of urban scenes. Experimental results show it is suitable for not only modern buildings but also classical architectures.