Locally adaptive bivariate shrinkage algorithm for image denoising based on Nonsubsampled Contourlet Transform

摘要

The Nonsubsampled Contourlet Transform (NSCT) is a new image representation approach that has sparser representation at both spatial and directional resolution as well as captures smooth contours in images. There are strong correlations between the parent and child coefficients of NSCT. Considering inter-scale and intra-scale dependency, in this paper, a method for image denoising in NSCT domain by using locally adapt bivariate shrinkage algorithm is proposed. This scheme achieved estimation results for images that are corrupted by additive Gaussian white noise (AGWN) and compares with NSCT-LAS, BivShrink and BLS-GSM. Experimental results show the proposed scheme can receive better denoising results.