Inverse halftoning based on HVS

The objective of this work is to reconstruct high quality graylevel images from bilevel halftone images, especially halftones with smooth dark regions, since human eyes are more sensitive to the noise in such area. According to the characteristic of human vision, a quantificational mapping between the halftone image and the contone image is built, then a preprocessor for the dark area is constructed. In order to get the contone image with good visual quality, a fast filtering method is used to refine the preprocessed halftone. The algorithm has advantages of both excellent reconstructed quality and fast speed.