Urban Environment Modeling from Mixed Airborne and Ground-Based Imagery

The modeling of urban environments has received ample attention over the past decades. Research has addressed the automatic derivation of urban geometry from a variety of data sources (imagery, laser, radar). This paper presents applied research of environment modeling techniques for defence and civil security applications. In our search for effective modeling work flows we pursue a path that exploits the combined value of airborne and ground-based imagery. The message of this paper is that this combination of source data is effective. This is demonstrated by presenting a case study for a small urban test area using stereo aerial imagery and ground-based panoramic photographs (socalled cycloramas). An algorithm is presented that automatically maps textures from the panoramic photographs onto the buildings. The algorithm works effectively, but depends on the availability of an accurate building geometry model and accurate registration of the ground-based imagery. Future work aims at automating the geometry extraction from the combined imagery sources and resolving occlusion problems intrinsic to ground-based imagery