Mixed Image Denoising Method of Non-local Means and Adaptive Bayesian Threshold Estimation in NSCT Domain

摘要

Image denoising is an important task inside the image processing area, a mixed image denoising method based on non-local means (NL-means) and adaptive bayesian threshold estimation in nonsubsampled contourlet transform (NSCT) is proposed. In this algorithm, first we remove the noise using NL-means method in spatial domain, then the denoised image using NL-means method is decomposed by NSCT into a low frequency subband and a set of multiscale and multidirectional high frequency subbands. The high frequency coefficients are estimated by the minimizing Bayesian risk. then the denoising image is gotten by performing the inverse NSCT to these estimated coefficents. Experimental results show that the proposed method indeed removes noise significantly and retains most image edges. The results compare favorably with the reported results in the recent denoising literature.