The nonsubsampled complex contourlet transform for image denoising

Abstract

As a remedy for pseudo-Gibbs phenomenon commonly existing in contourlet transform image denoising, a perfectly translation-invariant image denoising technique based on nonsubsampled complex contourlet transform (NCCT) is put forward in this paper. First, it use the dual-tree complex wavelet transform (DT-CWT) for multi-scale decomposition. Second, we apply nonsubsampled directional filter banks (NSDFB) to the DT-CWT coefficients, in such a way to maintain the anisotropy scaling law. The experiment results showed that this algorithm can get higher PSNR and the better visual.