More and more cities are looking for service providers able to deliver 3D city models in a short time. Airborne laser scanning techniques make it possible to acquire a three-dimensional point cloud leading almost instantaneously to Digital Surface Models (DSM), but these models are far from a topological 3D model needed by geographers or land surveyors. The aim of this paper is to present the pertinence and advantages of combining simultaneously the point cloud and the normalized DSM (nDSM) in the main steps of a building reconstruction approach. This approach has been implemented in order to exempt any additional data and to automate the process. The proposed workflow firstly extracts the off-terrain mask based on DSM. Then, it combines the point cloud and the DSM for extracting a building mask from the off-terrain. At last, based on the previously extracted building mask, the reconstruction of 3D flat roof models is carried out and analyzed. I.

In the field of disaster management the detection and classification of building damage play an important role. Airborne lidar data is very suitable as a basis for damage analyses because it can be acquired for large areas directly after a disaster. In building damage classification methods, plane surfaces extracted from post‐event lidar data are often used as one input. Various different algorithms exist for automatic plane detection from lidar data, of which two are presented in this paper and applied to lidar data of undamaged and damaged buildings. Finally, the suitability of these two algorithms for a more detailed building damage classification is studied and analysed.