Adaptive Wavelet Multi-thresholding for Image Denoising
BI Du-yan · 2005
Image denoising via wavelet transform is one success of wavelet applications, where the most important case is how to obtain the optimal threshold. This paper proposes an adaptive, data-driven multi-thresholding for image denoising based on different subbands and orientations according to visual performance. The thresholding is derived in a Bayesian framework, and the prior used on the wavelet coefficients is the generalized Gaussian distribution(GGD) which hass been widely used in image processing applications. Experiments show that this method is effective to image denoising. Comparing with Donoho’s Visu shrink and S.Grace Chang’s Bayes shrink, it not only improves the SNR(Singal-to-Noise Rate) and MSE(Minimizes the Mean Squared Error),but also makes denoised image more clear and fits to visual performance. Therefore, it have good performance both in objective and subjective.