Visual-Based error diffusion for printers

An approach for halftoning is presented that incorporates a printer model and also explicitly uses the human visual model. Conventional methods, such as clustered-dot screening or dispersed-dot screening, do not solve the gray-level distortion of printers and just implicitly use the eye as a lowpass filter. Error diffusion accounts for errors when processing subsequent pixels to minimize the overall mean-square errors. Recently developed model-based halftoning technique eliminates the effect of printer luminance distortion, but this method does not consider the filtering action of the eye, that is, some artifacts of standard error diffusion still exist when the printing resolution and view distance change. Another visual error diffusion method incorporates the human visual filter into error diffusion and results in improved noise characteristics and better resolution for structured image regions, but gray levels are still distorted. Experiments prove that human viewers judge the quality of a halftoning image based mainly on the region which exhibits the worst local error, and low-frequency distortions introduced by the halftoning process are responsible for more visually annoying artifacts in the halftone image than high-frequency distortion. Consequently, we adjust the correction factors of the feedback filter by local characteristics and adjust the dot patterns for some gray levels to minimize the visual blurred local error. Based on the human visual model, we obtain the visual-based error diffusion algorithm, and further we will also incorporate the printer model to correct the printing distortion. The artifacts connected with standard error diffusion are expected to be eliminated or decreased and therefore better print quality should be achieved. In addition to qualitative analysis, we also introduce a subjective evaluation of algorithms. The tests show that the algorithms developed here have improved the performance of error diffusion for printers