Median Filtering Forensics for JPEG
Compressed Images

Abstract

Image forensics deals with manipulations performed on the digital content. Median filter is popular noise removal operator whose detection has gained interest in forensic detection of image due to it’s non-linear nature and ability to hide traces of other image manipulations. Median filter introduces streaks (or) regions with constant intensity when applied on an image. In this work, Local Tetra Patterns (LTrP) is used for detection of median filter, which can capture the texture changes in a localized region caused by median filtering. LTrP(2 nd order) encodes the relationship between the referenced pixel and its neighbors, based on the directions that are calculated using the first order derivatives in vertical and horizontal directions. The J-Divergence is used for feature dimensionality reduction to decrease computational complexity. The experimental results shows that proposed method outperforms the state-of-arts in both detection accuracy and computational complexity.