Chinese Journal of Computers   Full Text
Title	Robust Detection of Region-Duplication Forgery in Digital Image
Authors	LUO Wei-Qi1),2) HUANG Ji-Wu1),2) QIU Guo-Ping3)
Address	1)(School of Information Science and Technology, Sun Yat-Sen University, Guangzhou 510275) 2)(Guangdong Key Laboratory of Information Security Technology, Guangzhou 510275) 3)(School of Computer Science, University of Nottingham, NG8 1BB, UK)
Year	2007
Issue	No.11(1998—2007)
Abstract &   Background	Abstract Region duplication forgery, in which a part of a digital image is copied and then pasted to another portion of the same image in order to conceal an important object in the scene, is one of the common image forgery techniques. This paper describes an efficient and robust algorithm for detecting and localizing this type of malicious tampering. The algorithm first divides an image into small overlapped blocks and then compares the similarity of these blocks, finally identifies possible duplicated regions using the main shift vector. The experimental results show that above method is more robust comparing with other existing algorithms and can successfully detect this type of tampering for images that have been subjected to various forms of post region duplication image processing, including, blurring, noise contamination, severe lossy compression, and a mixture of these processing operations. keywords blind image authentication; image forgery; detection of tampering; region duplication; robustness background Detection of Region-Duplication in digital image is in the field of passive/blind forensics, which is a new research area in information forensics. With the proliferation of digital cameras and computers, as well as software for image editing, the problem of digital image forgery is potentially very serious. Authentication of digital images becomes an important issue. Unlike the signature-based and watermark-based methods, which need to do some operations such as signature generated or watermark embedded in advance, the new forensic is passive and blind. It needs not any extra side information in detection. The method assumes that different imaging devices or processing etc. would introduce different inherent patterns into the output images. These underlying patterns are consistent in the original un-tampered images and would be altered after some kinds of manipulations. Therefore the patterns can be used as evidences for image source identification and alteration detection, i.e. the two main issues in passive forensics. Several research groups, e.g. Farid H et al. in Dartmouth College, Fridrich J et al. in University of Binghamton, Wu Min et al. in University of Maryland, Chang S F et al. in University of Columbia etc., have started to investigate the new technology and some works have been reported. However passive technology for image forensics is still in its infancy. There are many open issues. Region-Duplication is one of the common image forgery techniques. The attacker may perform some post processing attack after Region-Duplication operation, which makes the task of detecting forgery significantly harder. The key of the detection algorithm is the robustness against the post image processing, such as noise contamination, Lossy JPEG compression, blurring etc. Several researchers have developed methods for detecting this form of forgery. Fridrich analyzed the DCT coefficients for each block, while Farid employed principal component analysis(PCA) to capture the image blocks’ features. The authors’ method extracts the 7 low-frequency components for each overlapping block. All of the methods are based on block matching. The main difference of these methods is the choice of the features. From the experiments, the features of the authors’ methods are more robust to various post region duplication image processing operations comparing with the prior methods. This work is supported by the National Natural Science Foundation of China under grant No.90604008, the National Natural Science Funds for Distinguished Young Scholar under grant No.60325208 and the National Natural Science Foundation of Guangdong under grant No.04205407.