Factor Histogram based Forgery Localization in Double Compressed JPEG Images

Today convincing digital forgery can be created without master learning of image editing software. These fake pictures over exceptionally quick media may cause extreme results in the public arena. Passive digital image forensic is an area which uncovers these problems. Since JPEG compression deals with 8 × 8 DCT matrix it makes its own fingerprint which can be utilized to distinguish further forgeries in the picture. In this paper, we have proposed a technique which automatically locates forgery in the image based on histogram of DCT coefficient factors, called as factor histogram. When image undergoes aligned double compression this factor histogram shows peak at current quantization step as well as primary quantization step. Our algorithm searches for absence of such double maxima in block-wise factor histogram to identify tampered region. This method can find copy-move, copy-paste as well as pre-processed forgeries such as rotation and scaling.