Shearlet-based image denoising using bivariate model

Abstract

An adaptive Bayesian estimator for image denoising in shearlet domain is presented, where bivariate probability densities are used as the prior model of shearlet coefficients of images. The bivariate probability density function is proposed to model the statistical dependence between a coefficient and its parent and it is shown to fit very well to the observed noise-free histograms. Under this prior, a Bayesian shearlet estimator is derived by using the maximum a posterior (MAP) rule. Finally, a simulation is carried out to show the effectiveness of the new estimator. Experimental results show the proposed method can effectively reduce noise and remain edges, obtain better visual effect and higher PSNR.