A New Image Denoising and Enhancement Method Combining the Nonsubsampled Contourlet Transform and Improved Total Variation

摘要

Transform-based denoising methods are very popular in recent years. However, they often suffer from unwanted artifacts like pesudo-Gibbs phenomena. In this paper, we propose a new hybrid image denoising by combining the nonsubsumpled contourlet transform (NSCT) with improved total variation. First, an improved stark function which integrates noise reduction with feature enhancement is developed to nonlinearly shrink and stretch the NSCT coefficients. Then an improved Total variation is introduced to reduce the pseudo-Gibbs artifacts of the enhanced image which are caused by the elimination of small NSCT coefficients. Numerical experiments show that this approach improves the image quality by enhancing the shape of edges and important detailed features while suppressing noise in comparison to many well known methods.