Terrain Reconstruction from Multiple Views

Abstrac t . A two-stage approach is discussed for reconstructing a dense digital elevation model (DEM) of the terrain from multiple pre-calibrated images taken by distinct cameras at different time under various illumination. First, the terrain DEM and orthoimage are obtained by independent voxel-based reconstruction of the terrain points using simple relations between the corresponding image gray values. As distinct from other approaches, possible occlusions and changing shadow layouts are taken into account implicitly by evaluating a confidence of every reconstructed terrain point. Then, the reconstructed DEM is refined by excluding occlusions of more confident points by less confident ones and smoothed with due account of the confidence values. Experiments with RADIUS model-board images show that the final refined and smoothed DEM gives a feasible approximation to the desired terrain.