Boundary regularization and building reconstruction based on terrestrial laser scanning data

Digitization and modeling of city buildings is always one of the most important issues in Virtual 3D City. Traditional methods are usually expensive and time-consuming. Nowadays reconstruction methods based on LiDAR (Light Detection And Range) data have been widely used [1] due to its powerful capacity of data acquisition. Airborne Laser Scanning (ALS) data have got some promising results [2]. However the obvious disadvantage is that it can hardly get the profile information except for the roof of building, and the result is that people can not get complete building model. On the contrary, Terrestrial Laser Scanning (TLS) could get points cloud of buildings with high density and accuracy by multi-stations scanning for realistic 3D model reconstruction. The information includes not only building roofs, but also building profiles. In this paper, the authors use TLS data, and explore an automatic model reconstruction approach to establish digital building model. The registered cloud points from multiple scanning stations are taken as input parameters. Profiles information of building is extracted from raw points cloud by a segmentation algorithm. And a regular polyhedron model of building is finally presented. The experiment results show that the automatic reconstruction procedure and algorithms are easy to implement and can get reasonable building model after facade boundary regularization.