Image Denoising Based on Wavelet Domain Wiener Filtering

The parameters of traditional wavelet domain local Wiener filter are estimated from neighborhood,consisting of coefficients at adjacent spatial location and coefficients at adjacent scales.Because of the limited size of the neighborhood,the problem of the estimation accuracy arises.Aimed to resolve the problem,in this paper,analysis of the errors occurring in the traditional wavelet domain local Wiener filtering is presented,then according to the results of analysis,an improved method is proposed,which thresholds the wavelet coefficients by an appropriate threshold before the Wiener filtering.For the test images corrupted by noise with different levels,simulation results show that the improved method can effectively improve the performance of wavelet domain Wiener filtering,and the higher the noise levels,the more obvious the improvement.