An adaptive fast filtering algorithm is proposed for the filtering problem about gray-scale image is that polluted by salt and pepper noise of different degree. This algorithm protects image details by using different filtering schemes for pixels’ different character; and controls iteration number adaptively by means of possible noise density. In this way to remove salt and pepper noise with different density. In addition

this algorithm proposes a statistical collation recording method based on string to reduce the number of statistical collation

in this way the time cost of this algorithm’s one time filtering process equals to that of the standard median filtering algorithm. The experiment shows that this algorithm has strong adaptive ability

high calculating efficiency and can achiere good filtering results.