The work aims to put forward a multi-scale error diffusion halftone algorithm considering edge information, in order to solve the problems of tone darkening and edge blurring of the resulting halftone image caused by the traditional multi-scale error diffusion halftone algorithm. Firstly, a dynamic non-causal error diffusion filter was designed according to the states of the pixel points in the neighborhood of the pixel being processed currently. Then, the image edge information was extracted by means of the local average gray scale and spatial change rate of the image. Finally, the halftone processing sequence was determined by combining the edge and gray scale information. The visual effect of halftone processing results was uniform. The tone and edge structure information was well consistent with the original image. PSNR, WSNR and SSIM were respectively improved by 0.7-1.2 dB, 0.9-1.5 dB and 0.11-0.18. The proposed algorithm solves the problems of tone darkening and edge blurring of the multi-scale error diffusion algorithm, thus effectively improving the halftone output quality.