Pixel Based Image Forensic Technique for Copy-move Forgery Detection Using Auto Color Correlogram

摘要

The credibility of a digital image plays a generous role in blind Image forensics. We focus on detection of a typical kind of forgery, referred to as copy-move forgery, which is commonly practiced in the field of blind image forensics. In this kind of forgery as the forged region belongs to same image, it introduces inconsistency at pixel level in the tampered image. In the proposed work, we exploit a content based image retrieval feature extraction technique for detection of forgery. Auto Color Correlogram, which is a low complexity feature extraction technique, is employed to obtain feature vectors from the forged image. ACC is not used for detection of copy-move forgery in the previous detection schemes. The scheme is also successful in detecting forged region which is scaled or rotated on pasting, also effectively detects multiple region duplication within the image.