A probability model for adaptive image denoising based on the wavelet transform is proposed.The new model includes an interscale model and an intrascale model.Firstly
we use the interscale model to classify the coefficients into two classes:significant coefficients and insignificant coefficients.Secondly
the maximum a posteriori (MAP) estimator based on the intrascale model is used to restore the noisy wavelet image coefficients.The same model can be applied to the complex wavelet domain.Experimental results and analysis are given to demonstrate the validity and effectiveness of the proposed model.