Image Registration in Frequency and Spatial Domain

Abstract

In order to improve the registration accuracy and reduce the computational load of remote sensing image, a fast and robust method having two-stage is proposed in this paper. It has solved the problem that the computation complexity in the previous papers is large in the spatial domain and great errors exist in the frequency domain. Coarse estimation is done in the frequency domain using phase correlation algorithm. The refinement stage consists of extracting reliable feature points based on the modified SUSAN algorithm, and the feature point matching scheme according to the proposed method which combines pixel-based and feature-based methods. At last, the relation parameters between the pair of images are calculated. The experimental results show that the method is robust and has good efficiency.