A modified SFS algorithm based on stereo images for the three-dimension reconstruction of Urban buildings

Abstract

The blind areas of two adjacent aerial images are usually large in Urban Remote Sensing, in order to reconstruct three-dimensional (3D) information of the building in these blind areas, a new modified SFS algorithm based on stereo images is proposed. This SFS algorithm is utilized to reconstruct object surface gradient information (relative elevation) of buildings in the overlapping areas based on the grey information of the image. The absolute elevation information could be gained by the stereo images which had been already finished relative orientation and absolute orientation. The fitting transformational relationship is set up between the relative and absolute elevation information of the characteristic ground points in these overlapping areas with Least Square Method. The building 3D information in non-overlapping domain can be reconstructed by this transformational relation. From the experiment results, it is found that the stereo models reconstructed by the modified SFS algorithm based on stereo images are better and have lower distortions than those reconstructed by general SFS algorithm.