Improved Tampering Localization in Digital Image Forensics Based on Maximal Entropy Random Walk

Abstract— To use maximal entropy random walk on a graph for tampering localization in digital image forensics. Our approach serves as an additional post-pro-cessing step after conventional sliding-window analysis with a forensic detector. Strong localization property of this random walk will highlight important regions and attenuate the back-ground-even for noisy response maps. Our evaluation shows that the proposed method can significantly outperform both the commonly used threshold-based decision, and the recently proposed optimization-based approach with a Markovian prior. PRECISE localization of tampered image regions is one of the most challenging problems in digital image forensics. While many forensic features are known to differ between pristine and tampered images, their application for blind localization poses additional challenges. Typically, the localization capability is obtained by analysing the image in a sliding-window manner. Existing schemes include identification of local in consistencies of the photo response non-uniformity(PRNU)pattern < final year projects >