An evolutionary approach for image registration

Image registration plays an important role in many real-world applications such as remote sensing. A key issue of image registration is to find the hidden relationship between the input image and the reference image. In many cases, the hidden relationship is presented by a coordinate transformation matrix. Therefore, an image registration can be formulated as an optimization problem. In this paper, we propose to use evolutionary algorithms to optimize the transformation matrix. Instead of finding an optimal mapping between each pixel in the input and reference images, some local image features which are expressed as control points, are firstly extracted from the two images. An evolutionary algorithm is then applied to find the optimal mapping between the control points. Finally, the input image is registrated by the optimal transformation. The proposed approach is applied to some remote sensing images and the statistical results show that our approach is promising for dealing with image registration.