Color inverse halftoning method for scanned color images

A new color inverse halftoning method that converts a scanned color image, halftoned by the clustered-dot ordered dither, into a more natural continuous-tone image, is proposed by analyzing the Fourier spectrum of color image. A color channel separated from the color image makes three kinds of peaks in its Fourier spectrum: a channel peak, an interference peak, and a moire peak. The channel peak is formed by the repeated pattern of the channel halftone cells in the color channel, whereas the interference peak is made by the other channel halftone cells, due to imperfect separation of the color channel. The moire peak is the secondary peak formed by the interaction of the channel peak with the interference peak. A new smoothing mask for each color channel was designed to effectively remove not only the channel peak but also the interference peak, thereby making a smoother continuous-tone image. A block based moire extraction algorithm has also been developed to remove the moire peak adaptively, which can smooth the moire region in the color channel without blurring the other regions. Experiments show that the proposed method outperforms all the published results from other color inverse halftoning methods.