Security of copy-move forgery detection techniques

Copy-move forgery is a specific form of image tampering, where a part of a digital image is copied and pasted into a different part of the same image. There are many copy move forgery detection techniques, but their security has not been examined in detail. In this paper, we analyze the robustness and security of the detection techniques of Fridrich [1], Popescu [2], and Luo [3]. We show that all three schemes can successfully be defeated by targeted attacks, which are specifically tailored towards exploiting some characteristics of the schemes. Thus, with little effort an attacker can disguise a forged image as authentic even in presence of image forensic tools.