A Denoising Algorithm Baised on Nonsubsambled Contourlet Transform and Two-Dimensional Principle Component Analysis

Abstract

This paper proposes a novel image denoising algorithm which utilizes noise energy to perform image denoising based on Two-Dimensional Principal Component Analysis (TDPCA) in Nonsubsampled Contourlet Transform (NSCT) domain. The NSCT can capture the edges of natural images efficiently, and at the same time, it can get rid of the Gibbs effect. The noisy image can be decomposed by the NSCT into directional subbands. The TDPCA is then used to estimate the noise energy (local threshold) for the image blocks in high frequency subbands. The soft threshold shrinkage can hence be employed on the NSCT coefficients without estimating the noise variance. At last, the inverse NSCT is carried out on the modified coefficients to obtain the denoised image. The denoising algorithm is validated by numerical experiments on two typical images. Numerical results show that the proposed method is superior both in vision and in PSNR to former methods.