Abstract: The concepts of standard median filtering and effective median filtering are presented，and a median filter based on adaptive multiple scale noise detecting is proposed to restore images corrupted by salt-pepper impulse noise.The filter divides pixels of input image into effective signal class，impulsive noise class as well as constant region class.Then the pixels of various class are treated by different methods.Extensive experimental results demonstrate that the proposed algorithm performs significantly better than many existing decision-based methods，and the implementation of this algorithm is very simple.