HALFTONING WITH GRADIENT-BASED SELECTION OF DITHER MATRICES

Halftoning gray scale image data includes selection of one of plural dither matrices based on an analysis of local tonality gradient for each target pixel. Plural different but reciprocally compatible dither matrices are defined, and for each target pixel in the gray scale image data, a local gradient for tonality is calculated. One of the dither matrices is selected in accordance with the calculated gradient, and the target pixel is thresholded using the selected dither matrix. Enhanced dithering according to the invention improves smoothness of gradations present in the invention, diminishes the "banding" or "contouring" effect found in natural images, and improves visual appearance and stability of business graphics, particularly in cases where a complex image contains both natural and computer generated business graphics.