Impulse Noise Replacement With Adaptive Neighborhood Median Filtering

This paper presents an impulse noise replacement scheme based on an adaptive neighborhood median filtering (ANMF). Note that the window size used in the filtering process affects the contrast and smoothness in restored images. That is, larger windows applied in the filtering process result in a stronger smoothing effect and less contrast in restored images. On the other hand, a smaller window leads to a better contrast and less smoothness in restored images. Thus, three adaptive window expansion criteria are employed in the proposed ANMF schemes such that smaller windows are used in the replacement of noisy pixels. To justify the proposed ANFM schemes, three images are given where the salt and pepper noise with various densities are under study. The results indicate that the proposed ANFM schemes have better visual quality of restored images than those by [12], even though less peak signal-to-noise ratio is for the proposed ANFM in some cases of higher noise densities.