Tradeoffs Between Forensics and Anti-Forensics of Digital Images

Introduction

Due to the advancement and easy availability of digital editing soft wares, creating the forged contents is not a big deal, today. Because of this, one must verify the contents before using it. As a consequence, several forensic techniques have developed to detect various types of forgeries (Popescu & Farid, 2005; Farid, 2001; Kirchner & Fridrich, 2010; Farid, 2009). Staring from acquisition to fruition, image undergoes many steps and a forger can attack at any step. But, most image altering operations leave behind distinct, traceable “fingerprints” in the form of image alteration artifacts. As these fingerprints solely belong to that step/operation in which they get introduced, specific assessment to catch them must be designed. Such fingerprints are clues for forensic agents. They reveal the traces of fingerprints and present it to prove the presence of image altering operation. In order to study the flaws and limitations of current forensic tools and increase their strength with respect to current attacks, an intelligent forger must design techniques able to conceal the significant footprints left by typical image processing operation. In effort to conceal the footprints, distortion may get introduce, which results in poor image quality. Poor image quality is a vital parameter used by the forensic agents to detect forgeries. Thus, a trade-off between image quality and detectability gets introduced. If quality is preferred, footprints can be revealed and image alteration gets detected. On the contrast, if footprints are concealed, image gets distorted and thus image alteration can be detected. The main challenge in front of anti-forensic agents is to balance this tradeoff between image quality conceal ability of fingerprints and its un-detectability by forensic agents.

Compressing the multimedia signal for storage and transmission is quite common. Forgers also need to compresses their forgeries for the same purpose. During compression, a famous tradeoff between the compression ratio and distortion into the signal must be balanced. Therefore, in the case of compressing anti-forensically forged signal, compression ratio, image distortion and probability that the forgery is concealed are all linked. A forger must balance a tradeoff between all three of these quantities. In this paper, we review recent techniques proposed to reduce these tradeoffs and propose a framework for the same.