An efficient two-state switching median filter for the reduction of impulse noises with different distributions

Most of the median filters are designed to deal with salt and pepper noise or uniformly distributed impulse noise. A new switching median filter is proposed in this paper to restore images corrupted by either of the two. The proposed filter consists of two stages. The number of pixels at a noisy intensity is usually large and more likely to decrease after a standard median filtering process, so a noisy intensity identification scheme is used first to classify all intensities as uncorrupted intensities or noisy intensities. Then, further filtering scheme is chosen from two different switching median filters, according to the identification result. One only processes those pixels ay noisy intensities to avoid modification of noise free pixels when the identification scheme successes. The other takes a two stage noise detecting scheme to improve the filtering performance when the identification scheme fails at uniformly distributed impulse noise. Experiment results are given to show the superiority of the new proposed filter.