Abstract
One type of common printer banding artifact is a thin white horizontal line. This can result from a clogged ink jet nozzle, from paper-feed problems in electrophotographic printers, and from other causes. The halftone texture might be expected to mask the visibility of the line, so a line on a strong texture might be less visible than a line on a uniform field. To test this hypothesis, white lines were printed on fields with different texture contrasts, and we measured the threshold visibility for detecting the line.The line was printed as a reflectance increment in one of the pixel-rows of an error-diffusion pattern. The error diffusion-pattern was produced by applying the Floyd and Steinberg algorithm to a gray square. Two experiments were conducted. In one experiment, only the contrast of the halftone pattern was varied, and the subject was required to detect the pattern. In the second experiment, both the contrast of the halftone pattern and the contrast of the line were varied, and the subject was required to detect the line. The threshold for detecting the line was then plotted as a function of the strength of the halftone.It was found that the error-diffusion halftone pattern does not mask sensitivity to a line until the halftone pattern was many times above its own detection strength. Even when the halftone pattern was at the highest contrast that could be printed (about 20 times threshold contrast), the threshold amplitude for detecting the line was only doubled. Error diffusion is typically used because it reduces the visibility of the halftone texture. Since only very strong halftone textures provided any visual masking, the contrast of the error diffusion texture had to be very high to obtain any significant masking effect. This has the unfortunate implication that objectionable banding may only be hidden with even more objectionable halftone texture. Some of the implications for modeling band visibility will be discussed. It is important to note that there are previously described halftone patterns that can physically reduce the amplitude of the bands, and the present result does not run counter to these findings.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have