An image denoising method based on Markov-Chain Monte Carlo sampling with alterable direction and low rank approximation

Abstract

The proposed image denoising method investigates a novel similar block searching strategy based on non-local Markov-Chain Monte Carlo (MCMC) sampling with alterable direction. Firstly, observed image is decomposed with 2-D wavelet transform to obtain a series sub-band images in spatial Following, the similar matching block clusters of each sub-band image in spatial are obtained by taking the different sampling which obey different directional elliptical Gaussian distributions. The matrix of similar patches cluster is decomposed by singular value decomposition method, and the image noise is suppressed by applying the low rank structure from decomposing. The simulation results show that the proposed method outperforms the Block Method of 3-Dimension (BM3DJ and the Non-Local Means (NLM) methods in computational-complexity. The proposed method has a better performance in protecting image details compared with the NLM method, and has some advantages over the BM3D method in terms of visual quality.