Research On Automated Planar Information Extraction From3D LiDAR Point Cloud Of Different Platforms

Point cloud of different LiDAR platforms, because of their data acquisition efficiency and scanview complementarity, without any doubt, have become important data sources of theredimensional data acquisition method for construction of smart city. Starting directly from theoriginal there dimensional LiDAR point cloud data, this paper carries out a deep study on theautomatic extraction strategy of interested planar information, system framwork andimplementation algorithms on the basis of systematic summarizes and analysis of LiDARprinciples and data format. For the original point cloud of large scenes, an adaptive multi-gird sizefiltering algorithm is proposed, after automatic calculation of the initial grid size according todifferent dataset, the original LiDAR data of large scenes are filtered based on the local heightdifference of different grid. Thus, the original LiDAR point cloud can be filtered withoutinterpolation keeping the original precise information and improving the level of automation.Experiment results show that the proposed algorithm can seperate the groud and non-ground pointcloud well. For the non-groud point cloud, based on the analysis of existing segmentation methods,a segmentation method considering the local point density is proposed, comparative experimentsshow that the proposed method has better effect. For the detected planar cluster, the boundariesare extracted using-shape after coordinate transformation, experiments of airborne, terrestrialmobile and terrestrial static LiDAR data show that the method can effectively extracted planarinformation. By means of MFC and OpenGL, an automatic point cloud processing and visulationsystem is developed based on VS2010using VC++.