Adaptive Wavelet Shrinkage For Image Denoising Based On SURE Rule

Abstract

Based on the wavelet shrinkage denoising theory proposed by D.L. Donoho, a new thresholding function is presented in this paper, which is rather similar to hard thresholding one. However it has infinite-order continuous derivative. Compared to soft thresholding function, it can reserve better image details due to its "hard" characteristic. Moreover, the new one makes it possible to construct an adaptive algorithm for image denoising. By using the new thresholding function, a new adaptive shrinkage method is presented based on Stein's unbiased risk estimate (SURE). The two examples Lenna and Barbara are given. The results indicted that for image denoising application, the proposed method is very effective in adaptively finding the optimal solution in the least mean square error (LMSE) sense