A remote sensing image subpixel matching algorithm combined edge with grey

Remote-sensing image matching is one of the most important and difficult tasks in computer vision and digital photogrammetry. In this paper, a subpixel matching algorithm combining edge matching with grey-scale matching is presented in order to precisely and reliably match a pair of remote-sensing images taken by dissimilar sensors at different times. The algorithm contains three main processing steps. First, the edges of objects are extracted with a fast LOG (Laplacian-of-Gaussians) operator. Then, coarse matching is implemented by using the principle of invariant line moments. Finally, under the constraint of the above coarse matching, a subpixel location is directly produced with grey-based least-square matching (LSM). Experiments are presented which demonstrate that the desired purpose has been achieved.