Wavelet-based image denoising using NeighShrink and BiShrink threshold functions

This paper presents image-denoising methods performed within wavelet domain scheme by incorporating neighboring coefficients, namely NeighShrink (G.Y. Chen et al., 2004), and at the same time, denoising the image with bivariate shrinkage function. The idea of bivariate shrinkage function (BiShrink (L. Sendur and I.W. Selesnick, 2002)) is to model the signal based on MAP estimation approach. In fact, signal can also be bivariately modeled with MMSE estimator. The first method of this work is to incorporate the BiShrink with MMSE model, called here MMSE_BiShrink. In our second proposed method, we incorporate neighboring wavelet coefficients with BiShrink, which is called here MAP_NBShrink. Finally, we proposed the third approach by applying MMSE estimation method with NeighShrink and BiShrink, which we call here MMSE_NBShrink. Experimental results show that ours proposed methods, MMSE_BiShrink and MAP_NBShrink and MMSE_NBShrink, have better PSNR than NeighShrink and BiShrink, respectively.