Abstract: In order to preserve more details in an image while removing salt and pepper noise, an adaptive extremum and median filtering algorithm is proposed, according to the feature of salt and pepper noise. The size of the filtering window is adjusted automatically by detecting the presence of signal points in the window. Max-Min operator is used as the noise detector and a suitable threshold is chosen to identify the noises from the pixels which are in the center of the window and the gray value is equal to the maximum gray value or the minimum gray value of the window. Then filter outs the detected noise with the median value of the gray value of all signal points in the window, and outputs the signal points directly without changes. The gray value of the pixels in the center of window which exceeds the maximum window is replaced with the mean of the gray value of neighboring four pixels. Simulation results demonstrate the validity of the algorithm, especially when the density is high.

Key words: adaptive extremum and median filtering, noise detection, threshold