Wavelet-Domain Image Denoising Using Contextual Hidden Markov Tree Model

A fuzzy logic based wavelet-domain contextual hidden Markov tree(HMT) model is proposed for image denoising. This method uses fuzzy membership to establish a continuous hidden Markov random field (HMRF) for estimation of the distribution of the hidden state in each iteration. Using these updated data, the parameters of HMT are maximized. This algorithm introduces intrascale dependence into HMT train which can efficiently improve the expectation results of the HMT model. Experimental results are presented in the end of this paper.