Detection of Image Region Duplication through Digital Forensics

Digital images, being the major information carriers in today's digital world, act as the primary sources of evidence towards any event in the court of law as well as media and broadcast industries. Nonetheless, the relative simplicity of editing and manipulating digital images, have made their reliability and trustworthiness largely questionable. In fact, seeing is no more believing, due to the fact that in today's digital age, there is an expanding number of vindictively altered pictures. Exploiting the present-day vast availability of an extensive variety of efficient and easy–to–use image processing software and desktop tools, digital image manipulations by an adversary has become extremely common and simple. One of the major issues in crime scene investigation depicted in an image, is figuring out whether the image is genuine or doctored. This can be a critical assignment when images are utilized as fundamental proof to impact legal judgement, for instance, in the court of law. Digital Forensics is a branch of science that deals with the investigation, verification and recovery of digital data and evidences remnant in digital devices such as computers, digital cameras and video recorders. In this research, we deal with digital image forensics. We focus on a specific class of image forgery, viz. copy–move forgery or region duplication, and its subsequent detection and localization. In this type of forgery, one part of the image is copied and moved into another region of the same image, with the malicious intent of repeating or obscuring significant image objects of interest. This form of digital image forgery becomes all the more difficult to be detected, when the adversary performs additional intelligent attacks, such as edge blurring, brightness adjustment, noise enhancement of the forged regions, or geometric attacks such as rotation or re–scale of the forged regions before duplicating those. Majorly, this research focuses on the development and performance analysis of copy–move forgery detection and localization techniques, while addressing the additional challenge of detecting geometric and post-processing based attacks. In this dissertation, we present a survey of state–of–the–art copy–move forgery detection techniques, along with the motivation and objectives of this research. We present, efficient proposed techniques for region duplication detection (both block based and key–point based), and detailed experimental and comparison results pertaining to those. Finally, we conclude with significant future research directions