Wavelet transform and fuzzy reasoning based image fusion algorithm

Abstract

Image fusion based on wavelet transform is the most commonly used image fusion method, which fuses the source images’ information in wavelet domain according to some fusion rules. But because of the uncertainties of the source images’ contributions to the fused image, how to design a good fusion rule to integrate as much information as possible into the fused image becomes the most important problem. Both of the two main rules, the rule of selecting maximum absolute value and the combination rule of selecting and weighted averaging, ignore some useful information and are sensitive to noise. On the other hand, fuzzy reasoning is the best way to resolve uncertain problems, but it has not been used in the design of fusion rule. This paper proposed a new image fusion algorithm based on wavelet transform and fuzzy reasoning. After doing wavelet transform to source images, it computes the weight of each source image’s coefficients through fuzzy reasoning and then fuses the coefficients through weighted averaging with the computed weights to obtain a fused image. Using the mutual information and PSNR as criterions, experiment results demonstrated that the new method was more effective and robust than the traditional fusion methods based on wavelet transform, especially in some cases that source images are stained by noise.