Discrepancy-based digital halftoning: Automatic evaluation and optimization

Digital halftoning is a problem of computing a binary image approximating an input gray (or color) image. We consider two problems on digital halftoning: mathematical evaluation of a halftoning image and design of optimization-based halftoning algorithms. We propose an efficient automatic evaluation system of halftoning images by using quality evaluation functions based on discrepancy measures. Our experimental results on the evaluation system infer that the discrepancy corresponding to a regional error is a good evaluation measurement, and thus we design algorithms to reduce this discrepancy measure.