Combined Feature Extraction for Façade Reconstruction

Within the paper, the combined application of terrestrial image and LIDAR data for façade reconstruction is discussed. Existing 3D building models as they are available from airborne data collection are additionally integrated into the process. These given models provide a priori information, which efficiently supports both the georeferencing of the terrestrial data and the subsequent geometric refinement. Approximate orientation parameters for the terrestrial LIDAR measurements are provided by suitable low-cost components. Exact georeferencing is then realised by an automatic alignment to the building models, which are given in the required reference coordinate system. The automatic relative orientation of the terrestrial images is implemented by tie point matching. A modified version of this matching process is then used to align these images to the terrestrial LIDAR data, which were already georeferenced in the preceding step. After this fully automatic orientation process, the given 3D model is refined by the extraction of window structures from the LIDAR point clouds and façade images.