Abstract:A noise filtering algorithm was proposed for high-noise-ratio images corrupted by salt-pepper noise. In this algorithm, a pixel was identified as disperse noise (or edge pixel), clustered noise or no-edge pixel in an improved way based on gray absolute correlation degree. Then, disperse noise (or edge pixel) was smoothed using adaptive median filtering algorithm, which gave consideration to noise filtering and details keep togethor. And clustered noise was removed with the same value pixels within the neighbourhood area and replaced by mean value of the remainder pixels. Meanwhile, no-edge pixel was retained. Theoretical and experimental results show that the proposed algorithm performs effectivelly to high-noise-ratio salt-pepper noise, and it keeps details also.