A novel non-rigid registration method

Image registration is a fundamental step in image processing systems where there is a need to match two more images. Applications include motion detection, target recognition, video processing, and medical imaging. Although a vast number of methods have been proposed, it is not possible to claim that the problem is solved, especially when the deformation is locally variant and nonrigid. In this paper, we propose a registration method that is superior to other available algorithms, avoiding user interaction, and manual parameters, being noniterative (having a closed form solution), locally-variant, and allowing for sub-pixel registration. The method is successfully applied using computer simulations.