Properties of Jointly-Blue Noise Masks and Applications to Color Halftoning

Abstract

With the emergence of blue noise halftoning as a preferred halftoning technique, the issues of Moiré patterns in conventional ordered dither color halftoning are replaced by new issues concerning the quality of overlaid blue noise patterns. The goal of blue noise mask design in color halftoning is to generate a set of blue noise masks that produce high quality blue noise patterns whether they are used individually or jointly. We first address this question by inspecting various combinations of unstructured binary patterns, then explore the elementary properties of combinations of blue noise binary patterns. Based on the above analysis, an algorithm using filter techniques is proposed. This algorithm generates a set of jointly blue noise masks (JBNM) and each mask of the JBNM will be applied to one color plane. The JBNM possesses the properties such that each individual mask is blue noise, also the combination of the masks produce color images with blue noise characteristics, e.g., the dots from different color planes are mutually exclusive and maximally dispersed at highlight levels. Examples of halftoned color patches using JBNM and other schemes are illustrated. The evaluation demonstrates that JBNM can produce high quality color halftone patterns with little texture and graininess.