Resampling Halftone Images Using Interpolation and Error-Diffusion

Abstract

Halftoning schemes are developed for reserving the quality after transforming continuous tone images to binary images under human vision, and are widely used in printing and outputting. Modern halftoning schemes focus on transforming forward and backward the two kinds of images, but basic operations for halftone images like rescaling, blurring and sharpening are still insufficient. In this paper, a novel halftone image resampling scheme based on interpolation and error-diffusion is proposed. The proposed scheme can shrink and magnify halftone images to any desired scale and does not limit the source of the halftone images. To obtain the best performance, an evaluation tool is proposed to test various combinations of interpolation techniques and error-diffusion kernels. Due to the lack of evaluation research, the evaluation metric called light invariance is developed to evaluate the consistency before and after resampling. By the proposed scheme, resampling halftone images does not need transforming to and back from continuous-tone images anymore.