Statistical image modeling in the contourlet domain using contextual hidden Markov models

In this paper, a contourlet contextual hidden Markov model (C-CHMM) is established for modeling contourlet images by adapting a previous CHMM for wavelet images (W-CHMM). A mutual information based context design procedure is presented, through which a new context has been constructed. The C-CHMM is tested in a denoising application with promising results, which verifies its effectiveness. This new model is demonstrated to be a better model for contourlet images than the state of the art contourlet hidden Markov tree model. As a general image model, it also shows more potential than the baseline W-CHMM.