In this paper, we present an image denoising technique by applying Wiener filter in wavelet domain. The stationary wavelet transform (SWT) was used in our algorithm because it provides a shift-invariance property. The noisy image is first decomposed with SWT into N levels, which results in 3N + 1 wavelet coefficient subbands. The Wiener filter is then applied to each of 3N high-frequency subbands. The output image is finally obtained by inverse SWT. Experiments were carried out by simulation on several images, in comparison with the universal wavelet thresholding method. It was shown that our algorithm yields higher performance in terms of peak signal to noise ratio (PSNR) and visual quality.
展开▼
