Demons Registration Based On SIFT Detection

Image registration is an important aspect of image processing. This technology is rapid development in recent years. It is applied to many aspects, such as computer vision, medicine, military, remote sensing and so on. Especially, image registration has achieved important applications in medical research.At present, the theory and application of image registration has been studied widely and deeply inland and abroad. It has been a hot spot. Therefore, there are many methods of image registration which have important influence. For example, the image registration of maximum mutual information, the image registration of hybrid mutual information, image registration that based on wavelet resolution, Demons image registration, SIFT feature points matching image registration, image registration that based on genetic algorithm, B spline image registration and so on.In this paper, the main work is to propose a new image registration method by combining the Demons of image registration with SIFT feature points technique. The proposed new algorithm is feasible and efficient for image registration. By optimizing the objective function in Demons algorithm with quasi-Newton, the vector u of deformation can be obtained. It can see the merits of pictorial results very directly and intuitively by detecting the registration results with SIFT. This detection avoids a lot of computation of the similarity between two images which reflects the efficiency of the proposed algorithm. Experiments prove that the proposed algorithm is effective and feasible.