Classification and Change Detection in Mobile Mapping LiDAR Point Clouds

Abstract Creating 3D models of the static environment is an important task for advancement driver assistance systems and autonomous driving. In this work, a reference map created from Mobile Mapping “light detection ranging” (LiDAR) dataset. The data was obtained in 14 measurement runs March to October 2017 Hannover consists total about 15 billion points. point cloud are first segmented by region growing then processed random forest classification, which divides segments into five classes (“facade”, “pole”, “fence”, “traffic sign”, “vegetation”) three dynamic (“vehicle”, “bicycle”, “person”) with overall accuracy 94%. All objects entered voxel grid, compare different epochs directly. next step, classified voxels combined result visibility analysis. Therefore, we use ray tracing algorithm detect traversed differentiate between empty space occlusion. Each as suitable or not its object class occupation state during epochs. Thereby, avoid eliminate were occluded some (e.g. parts building tree). However, that only temporarily present connected objects, such scaffolds awnings on buildings, included map. Overall, combination classification subsequent entry grid provides good useful results can be updated including new data.