Hierarchical Dem Refinement Using Surface Parallax

We present a multi-resolution approach to update and refine coarse 3D models of urban environments from a sequence of intensity images using surface parallax. A coarse and potentially incomplete depth map of the scene obtained from a Digital Elevation Map (DEM) is used as a reference surface which is refined and updated using this approach. We first estimate the camera motion using the reference depth map. Using the estimated camera motion, at each level in the multi-resolution framework, motion of 3D points on the reference surface is compensated, and the residual flow field, which is an epipolar field, is estimated and used to refine the depth map at that level. At a coarse resolution, the difference between the reference depth and the true depth will be small, leading to a small parallax field. The refined depth map from the coarser level is then propagated to the finer level and is used as a reference depth map at that level. Thus, significant deviations of an available model from a true model can be handled using this approach.