Fusion of noisy images based statistical model in shearlet domain

Abstract

A new image fusion approach based on the modeling of shearlet coefficients with normal inverse gaussian model is proposed. The approach focus on the fusion of noisy images. Based on the statistical model and additive white gaussian noise, an subband adaptive shrinkage function is derived by using the maximum a posteriori rule. And then, the new scheme for shearlet-domain image fusion is proposed by incorporating the adaptive shrinkage rule into the fusion scheme. Experimental results show the proposed method perfor very well with noisy images, outperform other conventional methods.