Clustering based Impulse Detection and Rank-ordered Regularization for the Restoration of Impulse Corrupted Digital Images

The paper puts forward a new filtering algorithm for the restoration of digital images corrupted by impulse noise. The proposed filter includes a Hard-C means clustering stage in the impulse detection phase to facilitate pixel classification and an adaptive impulse filtering scheme in the restoration phase. The impulse detection phase avoids mis-classification of signals as impulses by identifying the high frequency image details from impulsive pixels. The adaptive restoration phase identifies suitable impulse remover from among the reliable signals of a valid neighbourhood. Subjective and objective assessments favour the filtering scheme of the proposed algorithm at all impulse noise levels than many top-ranking filters.