Some studies on detection and filtering algorithms for the removal of random valued impulse noise

Removing random valued impulse noise (RVIN) is a challenging task in corrupted images. This article aims to study some detection and filtering algorithms which remove RVIN in images. In addition to some state-of-the-art detection and filtering algorithms, a new detection technique, measures of dispersion (MOD) algorithm, for removing very high-density RVIN proposed by authors is also compared with existing methods. In the detection stage, rank order absolute difference, rank order logarithmic difference, adaptive switching median, triangle-based linear interpolation, and MOD algorithms are considered. Median filter, fuzzy switching median filter, and fuzzy switching weighted median filter are used for filtering followed by the detection algorithms. Comparative studies in terms of peak signal-to-noise ratio and structural similarity have been devised to evaluate the performance of various filtering schemes.