Colour Contribution For Stereo Image Matching

We examine the contribution of colour for the computation of depth maps from a stereo pair of colour aerial images. As a matter of fact we propose an original method based on fusion theory to compute dense and reliable depth maps. The fusion is directly applied on correlation curves obtained for each colour channel. In order to evaluate the quality of the elevation map, we compare our results to a 3D data reference. The results show that colour can improve the quality of the matching.