Large-scale 3D building reconstruction in LoD2 from ALS point clouds

Large-scale 3D building models are a fundamental data of many research and applications. The automatic reconstruction of these 3D models in LoD2 garners much attention and many automatic methods have been proposed. However, most existing solutions require multiple and complicated sub-steps for reconstructing the structure of a single building. Meanwhile, most of them have not been applied to large-scale reconstruction to better support the practical applications. Furthermore, some of them rely on the input point clouds with building classification information, thereby affecting their generalization. To resolve these issues, in this paper, we propose a workflow to fully automatically reconstruct large-scale 3D building models in LoD2. This workflow takes ALS point clouds as input and utilizes building footprints and digital terrain model (DTM) as assistance. LoD2 3D building models are reconstructed by a three-module pipeline: i) building and roof segmentation; ii) 3D roof reconstruction; iii) final top-down extrusion with terrain information. By proposing hybrid deep learning-based and rule-based methods for the first two modules, we ensure the accurate structure output of reconstruction results as much as possible. The experimental results on point clouds covering the whole city of Trondheim, Norway indicate that the proposed workflow can effectively reconstruct large-scale 3D building models in LoD2 with the acceptable RMSE.