An Efficient Restoration Algorithm for Images Corrupted with Salt and Pepper Noise

This paper presents a two-step restoration algorithm for impulse noise detection and removal. In the detection step, the pixel which is most likely corrupted by noise is detected according to its gray values. In the removal step, the proposed algorithm adaptively alters the filtering window size depending on the noise density. For a noisy pixel, if there exist one or more noise-free pixels in its window, the spatial correlation-based weighted mean filter will be applied to it by using only noise-free pixels. Otherwise, we use the median filter to correct the detection errors and remove noise. Naturally, the noise-free pixels are retained. Experimental results show that compared with the other filters, our algorithm can provide better performances in both quantitatively and visually.