Skip to Main Content
Motivated by the recent success of self-similarity in computer vision research, this paper proposed an approach for multimodal remote sensing image registration exploiting multiscale self-similarities (MSS) descriptor and coherent point sets analysis based on Gaussian mixture model (GMM) fitting. Rather than extracting sparse features for matching, we compute MSS descriptor at a regular grid. Point sets are selected according to their MSS descriptor similarity. Experimental results demonstrate the efficiency and the accuracy of the proposed technique for multimodal remote sensing image registration.