Multimodal image matching via double-angled transform and FFT-correlation

A multimodal image matching scheme is proposed. Firstly, gray-scale images in real field are converted into gradient images in complex field, and a double-angled transform is applied to assimilate the gradient orientations between the two images. Then, a generalized matching measure of image correlation in complex field is adopted to facilitate the treatment with geometric transform, and the matching surface is obtained efficiently by using the Fast Fourier Transform (FFT). The experiments verify that the proposed scheme is suited to matching multimodal images fast and robustly.