Median ltering detection based on variations and residuals in image forensics

To attain a robust feature vector for median ltering detection (MFD) in digital forgery images, this paper presents a short feature vector that is made up of three types of feature sets. The rst set is de ned by the variation to be the 3-D length in the gradient difference of the intensity values of the adjacent row and column line pairs in the image, respectively. The second set is de ned by the variation in the coefficient difference of the Fourier transform to be the 3-D length in the adjacent line pairs. The last set is de ned by the residual image between an image and its reconstructed image by the gradient based on solving Poisson's equation, which is also the 3-D length. Two of the sets are extracted in the spatial and spectral domains of an image, respectively, and the last set is extracted from the residual image. The totally formed 9-D feature vector is subsequently trained in the support vector machine classi er for MFD. In the experimental results of the proposed variation- and residual-based MFD scheme, the area under the curve is achieved closer to 1. Despite a short feature vector, the evaluation of the proposed MFD scheme is graded as Excellent (A)". In particular, the scheme detected good median ltering from the JPEG post-compression image for the cut-and-paste forgery image.

PDF

___

[1] Yuan H. Blind forensics of median ltering in digital images. IEEE T Inf Foren Sec 2011; 6-4: 1335-1345.
[2] Kang Hyeon RHEE. Median ltering detection using variation of adjacent line pairs for image forensic. In: IEEE 5th International Conference on Consumer Electronics; 2015; Berlin, Germany. pp. 103-107.
[3] Kang X, Stamm MC, Peng A, Liu KJR. Robust median ltering forensics using an autoregressive model. IEEE T Inf Foren Sec 2013; 8: 1456-1468.
[4] Chen C, Ni J, Huang J. Blind detection of median ltering in digital images: a difference domain based approach. IEEE T Image Process 2013; 22: 4699-4710.
[5] Kirchner M, Fridrich J. On detection of median ltering in digital images. SPIE Media Forensics and Security 2010; 2: 1-12.
[6] Zhang Y, Li S, Wang S, Shi YQ. Revealing the traces of median ltering using high-order local ternary patterns. IEEE Signal Proc Let 2014; 21: 275-280.
[7] Gui X, Li X, Qi W, Yang B. Blind median ltering detection based on histogram features. In: Annual Summit of the Asia-Paci c Signal and Information Processing Association; 9{12 December 2014; Siem Reap, Cambodia. pp. 1-4.
[8] Ke Y, Qin F, Min W, Zhang Q. An efficient blind detection algorithm of median ltered image. International Journal of Hybrid Information Technology 2015; 8: 181-192.
[9] Kay SM. Modern Spectral Estimation: Theory and Application. Englewood Cliffs, NJ, USA: Prentice Hall, 1998.
[10] Fattal R, Lischinski D, Werman M. Gradient domain high dynamic range compression. ACM T Graphic 2002; 21: 249-256.
[11] Raskar R, Tan KH, Feris R, Yu J, Turk M. Non-Photorealistic Camera: Depth Edge Detection and Stylized Rendering Using Multi-Flash Imaging. Mitsubishi TR2006-107. Tokyo, Japan: Mitsubishi, 2006.
[12] Chang CC, Lin CJ. LIBSVM: a library for support vector machines. ACM Transactions on Intelligent Systems and Technology 2011; 2: 1-27.
[13] Schaefer G, Stich M. UCID-an uncompressed color image database. In: Proceedings of Storage and Retrieval Methods and Applications for Multimedia Conference; 18 January 2004; San Jose, CA, USA. Bellingham, WA, USA: SPIE. pp. 472-480.
[14] Cao G , Zhao Y , Ni R , Yu L, Tian H. Forensic detection of median ltering in digital images. In: IEEE International Conference on Multimedia and Expo; 19{23 July 2010; Singapore. pp. 89-94.
[15] Xu J, Ling Y, Zheng X. Forensic detection of Gaussian low-pass ltering in digital images. In: 8th International Conference on Signal and Image Processing; 14{16 October 2015; Shenyang, China. pp. 819-823.
[16] Kang Hyeon RHEE. Median ltering detection using variation of neighboring line pairs for image forensics. Journal of Electronic Imaging 2016; 25: 1-13.