Bayesshrink based Wiener filter in wavelet domain

Empirical designed wavelet domain Wiener filters such as WienerChop have superior performance over other denoising algorithms using wavelet thresholding. An effective way to improve the denoising performance of WienerChop algorithm lies on the estimation precision of the expected signal. This paper proposed an improved WienerChop in which the Bayesian based wavelet thresholding denoising technique is firstly adopted to estimate the desired signal in the first wavelet domain to ensure the better estimation. Furthermore the iterative technique is also used by selecting multiple wavelet bases to capture some signal characteristics, which exist in the sparsity of the signal representation by wavelet transform. Theoretical analysis and simulation results show that our two methods, improved WienerChop and iterative WienerChop, outperform the traditional WienerChop algorithm in terms of PSNR and MSE.