Relation-Constrained 3D Reconstruction of Buildings in Metropolitan Areas from Photogrammetric Point Clouds

The complexity and variety of buildings the defects point cloud data are main challenges faced by 3D urban reconstruction from clouds, especially in metropolitan areas. In this paper, we developed a method that embeds multiple relations into procedural modelling process for automatic photogrammetric clouds. First, hybrid tree constructive solid geometry boundary representation (CSG-BRep) was built to decompose building bounding space polyhedral cells based on geometric-relation constraints. approximate shapes were then selected topological-relation constraints geometric models generated using reconstructing CSG-BRep tree. Finally, different parts retrieved trees, specific surface types recognized convert City Geography Markup Language (CityGML) format. clouds 105 area Hong Kong used evaluate performance proposed method. Compared with two existing methods, performed best terms robustness, regularity, topological correctness. CityGML enriched semantic information also compared manually digitized ground truth, high level consistency between results suggested produced will be useful smart city applications.