Point Cloud Reconstruction of Textureless Regions with Topology Constraints between Corresponding Points

Point clouds are generated with structure from motion/multi-view stereo (SfM/MVS) or laser scanning for building information modeling (BIM). Although the SfM/MVS can generate dense point clouds, it is not easy to reconstruct point clouds in texture-less regions because the SfM/MVS is based on feature point-based image matching. Thus, we propose a methodology to generate point clouds of all pixels in texture-less regions with the topology constraints among corresponding images.In this study, we use the back projection of point clouds to understand the topology of the corresponding points to improve the performance of corresponding points estimation and mismatching rejection with the epipolar constraints. We conducted an experiment onthe3D modeling of a metal bridge. Through our experiment, we confirmed that our methodology can reconstruct point clouds, even if targeted images containtexture-less regions.