Method of processing 3D sensor data to provide terrain segmentation

摘要

A method of processing 3D sensor data to provide terrain segmentation comprises processing the 3D data so as to analyse the positions of sets returns from the terrain in the horizontal plane. Returns from an idealised flat terrain form a series of scanlines (k-1, k, k+1) in the horizontal plane. A return (54) having a significant displacement in the horizontal plane in comparison to other returns (52, 56) in the same scan line is classified as a non-ground return corresponding to an elevated object (58). The method provides more reliable segmentation of ground areas than is possible with prior art techniques, and requires fewer processing resources because the vertical (height) positions of returns are not required to be processed. The method also has the advantage that small elevated objects having a height less than the range resolution of the 3D sensor may be identified.