The Application Of Nonlinear To Inverse Error Diffusion

Two inverse halftoning algorithms for error diffusion are proposed by nonlinear theory in this thesis. The influence of the property of human visual system and color space model on inverse halftoning algorithms is considered and the characteristics of noise distribution in color error-diffused halftone images are analyzed. It is the defect that general linear lowpass filtering of the noise leads to excessive blurring of the boundaries, so a color inverse halftoning algorithm is designed, which is based on edge-preserving smoothing model. Experiment shows that the algorithm has the efficient edge-preserving ability while smoothing the halftone patterns and excellent PSNR performance. For gray error diffusion image, a new algorithm of inverse halftoning by Genetic Algorithm is proposed. Experimental results demonstrate that this algorithm is practical and effective.