Registration of Mars remote sensing images under the crater constraint

Abstract

Registration of Mars remote sensing images is vital to jointly exploit and integrate the data from various sensors and periods, which is difficult because the lack of textural information in Mars images. We proposed a RANSAC-based iterative method, under the crater constraint, for affine invariant feature matching. The novelty of this method is, in automatic way, to find an optimal threshold for this RANSAC-based iterations. With this proposed method, a computer-automated process is developed to identify the appropriate threshold of RANSAC when using the iterative affine invariant feature matching method. Furthermore, this threshold can maximize the retention of correct matches and the rejection of all incorrect matches. The experimental results demonstrate successful automatic registration of Mars remote sensing images.