3D building reconstruction from LiDAR data based on Delaunay TIN approach

With a great progress in the research on digital city and related commercial applications, the construction of 3D city models has been matured gradually as a research topic in recent years. When constructing 3D city models on large scale, it is inevitable to have available reconstruction tools that offer a high level of automation and produce valid models reliably within the required accuracy. In this paper, a new method for the automatic generation of 3D building models from directly observed point cloud captured by airborne LiDAR system is presented. At first, Delaunay TIN model is constructed based on the building points. Then an edge length ratio based trace algorithm is used for refining the building's boundary. And the extracted outlines are regularized applying the regression analysis to obtain the better geometric quality. Then the same plane points set are clustered via those points in a triangle facet having the similar normal vector value, which determines the roof structures. Finally, with the regular outlines and roof structures information, the building models are constructed. The data in the study area provided by www.terrasolid.fi are used to test the approach. The experimental results show that this proposed approach is able to construct building models effectively.