Image restoration from corrupted halftone image using the statistical mechanical iterative method

In order to reconstruct a grayscale image from a noisy halftone image converted by the dither method, we construct a method of image restoration using the epsiv-filter and the statistical mechanical iterative method due to the Bethe approximation established in statistical physics to approximate the thermodynamics of magnetic materials. Then, using the numerical simulation for a 256-level standard image ldquoLenardquo with 512 times 512 pixels, we clarify that our method is successful in image restoration more accurately than the conventional filter, if we set the parameters appropriately.