An efficient LOF-based long-range correlation filter for the restoration of salt and pepper impulse corrupted digital images
Justin Varghese, Saudia Subash, Mohamed Samiulla Khan, Krishnan Nallaperumal, Bijoy Babu, Mohammed Ramadan SAADİ · TURKISH JOURNAL OF ELECTRICAL ENGINEERING & COMPUTER SCIENCES · 2016
The paper proposes an adaptive long-range correlation-based filter operator for the restoration of impulse corrupted digital images. The impulse detection scheme of the proposed algorithm incorporates the local outlier factor (LOF) to avoid the misclassification of uncorrupted pixels as noise. The restoration algorithm uses the local and remote neighborhood of the same size to find structural similarity among pixels for ensuring better replacement of detected impulses. The domain of the remote window is limited around the neighborhood of the impulsive pixel under concern for maintaining image details and thereby producing a high quality restored image. For replacing impulses, the filter uses a reference image and information about the corruption/purity status of the pixels in the image to determine the most correlated uncorrupted pixel from the remote neighborhood. Experimental results show that the proposed filter is capable of producing better results than the comparative filters in terms of subjective and objective metrics.