To reduce the influence of salt & pepper noise in images on subsequent image processing, a high-density salt & pepper noise filtering algorithm based on noise detection is proposed for high-density noise polluted images in this paper. Considering that potential function has the ability of edge-preserving, detailed image information is protected in a maximum scale and high-density noise polluted images can be restored. Experiment results show that the filtering algorithm presented in this article has high selfadaptability and good filtering effect.