Facades modeling from a ground-view video with map constraints

We present a 3D modeling approach to reconstruct urban facades with map constraints. Our main constribution is to use GIS data to filter the points cloud to improve the reconstruction process. The method's input is a ground-view video sequence captured by a common webcam. We extract original feature points cloud by using structure-from-motion technique. First, the footprint from GIS enables us to achieve filtering process, which is a key step to generate a quality reconstruction result in our method. Secondly, our method reconstructs the final facades based on the footprint. We show a simple and accurate facades modeling system to reconstruct building facades.