Digital Multitoning Using Gray Level Separation

Abstract

Digital halftoning has, traditionally, been thought of as a bilevel quantization algorithm that converts continuous tone images to images composed exclusively of “on” and “off” pixels. With recent advancements in printing technologies, ink jet and laser printers are now capable of producing dots with more than two intensity levels. These advancements have led to halftoning research on multitoning algorithms or halftoning with more than two levels. An early example of multitoning is the Floyd and Steinberg error diffusion algorithm with an N-level quantizer replacing the conventional binary quantizer. A major problem associated with this approach is the introduction of unwanted texture near the intermediate gray levels in the printed image. A possible solution to this problem is the redistribution of the intermediate gray and black pixels near the printable gray levels. In this article, a novel multitoning algorithm is introduced that can control the amount of printable gray level pixels, represents a particular shade of gray, by using gray level transformation. In this method, the input image is decomposed into the printable gray scale images, by using a set of gray level transformation curves, each channel is halftoned using conventional bilevel error diffusion algorithm, in a correlated way, and the halftoned channels are then recombined to get the final halftoned image. For certain gray level curve specifications, elimination of the undesirable banding artifacts has been achieved, near the intermediate gray levels, in the output. The suggested method is mean preserving and has a very little computational overhead involved as compared to the conventional error diffusion.