Non Blind Image Restoration Using Hidden Markov Modeling and Minimum Entropy Approach

In this paper a new method based on HMM (Hidden Markov Models) is presented for image restoration. We assume the image is treated by a finite impulse response (FIR) channel where the image is modelled as a Markov process. The superior property of this new method is the detection of the most likely gray level for each pixel based on the probabilities defined for both the noisy blurred observations and the original image. We also propose a new method to find the more structured region of the image based on the minimum entropy approach. Performance of the proposed algorithm is illustrated through simulations employing images blurred with different point spread functions. Comparison between the introduced method and the other proposed methods shows the superiority of HMM method specially for large spreading blurs.