A research on 3D reconstruction of building rooftop models from LiDAR data and orthophoto

A new method for 3D reconstruction of building rooftop models from LiDAR data and orthophto is proposed. The main contributions of this method include a triangular cluster - based algorithm for rooftop patch segmentation and a strategy to extract ridge lines of rooftop patches by the integration of LiDAR data and orthophoto. The process consists of a consequence of three steps. Firstly, a Delaunay triangular network is constructed from LiDAR data and a triangular cluster - based segmentation is executed to separate rooftop patches from this triangular network. Ridge lines of rooftops are then extracted by overlaying the extracted lines from orthophoto and the segmented rooftop patches. Finally, 3D rooftop models are reconstructed with the segmented rooftop patches and the extracted ridge lines. The experiment demonstrates that this proposed method can construct 3D building rooftop models with high correctness, completeness and geometric accuracy. For the reconstructed 3D rooftop models in test area 1 and 2, the correctness are 86%, 90%, respectively; the completeness are 85%, 88%, respectively; the positioning precision are 0.39m, 0.31m, respectively.