Impulse noise detection and correction by neighborhood switching median filter

In this paper, a first order neighborhood switching median filter algorithm is proposed to restore of gray scale images that are highly corrupted by salt and pepper noise. The proposed method removes only the noisy pixels by considering the first order neighborhood pixels. This proposed algorithm comprises two stages. The first stage describes the detection of noisy pixels by dividing the window into three clusters and in the second stage it considers only the noise free first order neighborhood pixels of the detected noisy pixel for the removal of noise. If all the first order neighborhood pixels of the detected noisy pixel are having gray value of 0 and 255, then the noisy pixel is replaced by the average values of those neighborhood pixels including it. When all the pixels in the first order neighborhood of the detected noisy pixel are not having gray value of 0 and 255, then the noisy pixel is replaced by median value of those neighboring pixels which values are not 0 or 255. The proposed algorithm out performs the performance of standard median filter, peer group based filter and boundary discriminative noise detective switching median filter.