During image information acquisition and transmission,the images are usually corrupted by impulse noise.To remove effectively high density impulse noise,an algorithm for removing noise based on a median-mean filter is proposed.According to the characteristics of impulse noise,a simple noise detection operator is set,and then adaptive filtering window is set according to the results of noise detection,meanwhile,the median and mean filters are selected according to the noise density.To keep effectively the original image information,the noise-free pixels are not filtered.Simulation results show that the proposed median-mean filter algorithm can remove high density impulse noise effectively.It also keep the original image information well,and have shorter filtering time.