THE ASCENDER SYSTEM - AUTOMATED SITE MODELING FROM MULTIPLE AERIAL IMAGES

The Ascender system acquires, extends, and refines 3D geometric site m odels from calibrated aerial imagery. To acquire a new site model, an automated building detector is run on one image to hypothesize potenti al building rooftops. Supporting evidence is located in other images v ia epipolar line segment matching in constrained search regions. The p recise 3D shape and location of each building is then determined by mu ltiimage triangulation under geometric constraints of 3D orthogonality , parallelness, colinearity, and coplanarity of lines and surfaces. Pr ojective mapping of image intensity information onto these polyhedral building models results in a realistic site model that can be rendered using virtual ''fly-through'' graphics. As new images of the site bec ome available, model extension and refinement procedures are performed to add previously unseen buildings and to improve the geometric accur acy of the existing 3D building models. In this way, the system gradua lly accumulates evidence over time to make the site model more complet e and more accurate. An extensive performance evaluation of component algorithms and the full system has been carried out. Two-dimensional b uilding detection accuracy, as well as accuracy of the three-dimension al building reconstruction, are presented for a representative data se t. (C) 1998 Academic Press.