关键词: 数字半调, 误差分散, HIST滤波器, 平稳小波变换

Abstract: By studying the filtering based error diffused inverse halftoning algorithms，a multiscale framework for inverse halftoning is proposed.With the framework and stationary wavelet transform，a new inverse halftoning method is presented.Firstly，halftone images are decomposited into three scales using stationary wavelet transform.By analyzing the frequency characteristics of each scale，construcing HIST filter，and considering the correlations between inter-scales，different schemes are made to remove the noise.Experiments show that the framework is right and the proposed algorithm is able to achieve an inverse halftoning image with good visual quality，it has a better edge-preserving ability and peak signal-to-noise ratio is increased.

Key words: digital halftone, error diffusion, HIST filter, stationary wavelet transform