Iterated conditional modes for inverse halftoning

In this paper, we present an approach that uses the iterated conditional modes (ICM) for stochastic inverse halftoning. The proposed ICM algorithms require only local computation and are applicable to any type of the MRF model used for the original gray-level images. The ICM estimation algorithms for inverse dithering and inverse error diffusion (ED) are respectively proposed. The algorithms are to find optimal solution completely in the valid image space, so the ICM approach prevents the possible oscillation in some kinds of two-phase descent-projection approach. The ICM approach is a better alternative to the latter approach. Experimental results are given to show the success of our ICM algorithms for inverse halftoning.