Image registration based on feature extraction and voting strategy

In this paper, we develop a coarse-to-fine method for the registration of the reference and target images, involving the 2D rigid transformation. The proposed method consists of four steps: utilizing Harris corner detector to extract the feature points in the reference and target images; corresponding these points by a novel feature points match algorithm based on voting strategy, which efficiently eliminates the false and unstable matches; estimating the transformation parameters with the corresponded feature points; refining the transformation parameters with the window-based method. Experimental results have demonstrated that the proposed method is able to achieve high accuracy and robustness with low computational burden.