Image Fusion Algorithms Based On Multi-scale Analysis

Image fusion is used to fuse two pictures or more to one new by appropriate algorithms.Multi-scale analysis comes from the simulation of human eyes'perception process in research on Computer Vision. And it is widely used in image fusion fields due to such good properties as localization. Images can be decomposed to several different scale parts, the coefficients of different scale roughly represent different feature of source image. Then proper fusion rules will be chosen to fuse the coefficients perfectly according to their characters. This targeted method improves the quality of the fused image.The main work of this paper is research algorithms based on Wavelet Transform and Curvelet Transform theory. In this paper many groups of different experiments used wavelet transform and curvelet transform with different fusion rules are realized, and the fused images are evaluated subjectively and objectively. Then we analysis the character of multi-focus images fusion, and this paper puts one improved method based on region average gradient used curvelet transform. We first process the source images by curvelet transform, and then fused the coefficients of different scale; the coarse scales are taken average while the fine scales are used improved region average gradient directed method, at last reconstruct the coefficient. Experiments show that the fusion rules this paper puts based on region average gradient are more effective than traditional ones, and the fused images are more closed to reference images.