Image denoising via doubly Wiener filtering with adaptive directional windows and Mean Shift algorithm in wavelet domain

An image is often corrupted by noise in its acquisition or transmission. The goal of denoising is to remove the noise while retaining as much as possible the important signal features. In this paper, we propose a doubly local Wiener filtering method using adaptive directional windows and Mean Shift algorithm, in which the Mean Shift algorithm is first used to naturally segment the image into regions of similar content, and then the adaptive directional windows which can change shape according to the different regions, are used to estimate the signal variances of noisy wavelet, finally the doubly local Wiener filtering is used to denoise the observed image. Simulations demonstrate this method substantially outperforms the original algorithm.