Image registration is one of the crucial steps in the fusion processing of image. Image registration, especially for remote sensing image, requires intensive computational efforts while the registration precision is not satisfying generally. This paper introduces a novel image registration method based on the wavelet with excellent performance. The method reduces the data to search via decomposing the image to several effectual subimages by the wavelet, then reduces the searching space through multi-resolution analysis, and refine the solution using an iterative refinement algorithm finally. Experimental results of remote sensing and optical images show that the proposed algorithm outperforms conventional approaches in terms of both computational effort and precision.