Skip to Main Content
In this paper, we describe a novel error diffusion scheme for higher halftone quality with less visual artifacts. The proposed algorithm improves mid-tone quality of error diffusion significantly by diffusing the error along a jumping scanning path. The algorithm calculates the accumulative error to determine the point where breaks up the scanning path. A cost function is developed to search the optimal threshold of the accumulative error. The cost function helps to bring the Fourier spectra as close as possible to the corresponding "blue noise" spectra, and minimize the anisotropy. Experiments show that compared with existing error diffusion algorithms that use continuous scanning path, the new algorithm can remove visual anomalies more effectively. Fourier analysis of experimental results further supports this conclusion.