Automated Reconstruction of 3D City Models from Laser Scans and Camera Images
7I will present a fast approach to automated generation of textured 3D city models with both high details at ground level and complete coverage for birdΓÇÖs-eye view. A ground-based facade model is acquired by driving a vehicle equipped with two 2D laser scanners and a digital camera under normal traffic conditions on public roads. One scanner is mounted horizontally and is used to determine the approximate component of relative motion along the movement of the acquisition vehicle via scan matching; the obtained relative motion estimates are concatenated to form an initial path. Assuming that features such as buildings are visible from both ground-based and airborne view, this initial path is globally corrected by Monte-Carlo Localization techniques using an aerial photograph or a Digital Surface Model as a global map. The second scanner is mounted vertically and is used to capture the 3D shape of the building facades. Applying a series of automated processing steps, a exture-mapped 3D facade model is reconstructed from the vertical laser scans and the camera images. In order to obtain an airborne model containing the roof and terrain shape complementary to the facade model, a Digital Surface Model is created from airborne laser scans, then triangulated, and finally texture-mapped with aerial imagery. Finally, the facade model and the airborne model are fused to one single model usable for both walk- and fly-thrus. The developed algorithms are evaluated on a large data set acquired in downtown Berkeley. Example models can be found at http://www-video.eecs.berkeley.edu/~frueh/3d/