A Method for Building Model Reconstruction Based on Multi-source Feature Fusion

Taking laser point cloud data and multi-view airborne oblique images as studying objects, an approach by integrating multi-source feature from airborne point cloud, ground laser point cloud and multiview airborne oblique images is proposed for building model reconstruction. Based on the combination of planar surface from point cloud segmentation and edge line extracted from images, the shape of the roof and façade model is refined utilizing the image lines as boundary constraints.Then the reconstruction process is implemented through the automatic topological optimization, necessary interactive editing and the final texture mapping. Compared with the manual and automatic reconstruction methods, the experiment result shows that this method can effectively improve the reliability of 3D model reconstruction of urban buildings, and provides a feasible solution for detailed building reconstruction using multi-source data.