Abstract
A recently developed mathematical analysis of error diffusion by Weissbach et al.1 describes the output spectrum of the error diffusion algorithm in terms of a linear filter equation. Their analysis showed that the output spectrum is given by the sum of the input spectrum and a high-pass filtered error spectrum. This analysis has been extended to error diffusion implemented on a serpentine raster. A serpentine raster is one in which the direction of processing is reversed for every scanline. The rationale behind this method is that the repeated reversal of the processing direction eliminates the directional artifacts in the output image that are caused by the asymmetrical distribution of errors in the algorithm. The new analysis of error diffusion on the serpentine raster shows that the nonsymmetric parts of the error spectrum are not eliminated but instead are reduced in magnitude and shifted to half the scanning frequency.
© 1991 Optical Society of America
PDF ArticleMore Like This
David Mendlovic, Hanni Inbar (Pinhas), and Emanuel Marom
WO2 OSA Annual Meeting (FIO) 1991
Barry L. Shoop and Joseph W. Goodman
ThHH1 OSA Annual Meeting (FIO) 1991
Reiner Eschbach and Zhigang Fan
TuRR5 OSA Annual Meeting (FIO) 1991