數位相機愈來愈普遍，除此之外，一些影像編輯軟體也愈做愈精細，使得數位相片雖然容易取得，卻也容易經由人為合成修改，而且這些人為的修改，常常是人眼無法分辨的。因此，數位鑑識(Digital Forensics)在現今是一個很重要的研究主題，而相機來源辨識(camera source identification)在數位鑑識的領域中是個特殊的議題。在本篇論文中，我們試著去分析可用於相機來源辨識的特徵。彩色內插演算法(color demosaicking algorithm)中的內插係數(interpolation coefficients)已被認為是描述不同相機的有效特徵。然而，在影像的不同區域所使用的內插係數可能不同，因此，我們提出將影像分離成平滑區(smooth region)及非平滑區(non-smooth region)，並針對平滑區及非平滑區修改EM演算法(EM algorithm)，修改的EM演算法將結合分離平滑區及非平滑區的結果來提升辨識結果。我們所修改的EM演算法所產生的波峰圖(peak map)是另一個在相機來源辨識上重要的特徵，我們在實驗的部份，會用不同的波峰表示法(peak representation)來測試相機來源辨識的結果。除此之外，我們會用主成分分析(principal component analysis)來達到降低維度(dimension reduction)的目的，將特徵維度(feature dimension)減少可加快學習(training)的時間。將我們取出來的特徵利用SVM來決定每張照片的相機來源。經由我們實驗說明我們所提出將影像分成平滑區及非平滑區，可以提升相機來源辨識的正確率；另外，我們所使用的波峰表示法也能有效的辨識相機來源。

Digital forensics is now an important research topic, and camera source identification is a special issue in digital forensics. In this work, we try to analyze useful features on camera source identification. Interpolation coefficients of color demosaicking algorithm have been proved to be an effective feature to characterize cameras. Here, we address smooth and non-smooth separation issue while different regions in the image may use different interpolation coefficients. The modified EM algorithm will combine the smooth and non-smooth separation result to improve the identification result. Peak map produced by our modified EM algorithm is another characteristic on camera source identification. We will use different peak representation to test the identification results on each case. Moreover, we use principal component analysis (PCA) for dimension reduction which will speed up the training time. Finally, we use a support vector machine (SVM) to determine the source of each digital image.

[1] S. Bayram, H.T. Sencar, N. Memon, and I. Avcibas, “Source camera identification based on CFA interpolation,” Proc. of International Conference on Image Processing, 2005.
[2] S. Bayram, H.T. Sencar, N. Memon, “Improvements on source camera-model identification based on CFA interpolation,” Proc. of the WG 11.9 International Conference on Digital Forensics, 2006.
[3] A.C. Popescu and H. Farid, “Exposing digital forgeries by detecting traces of resampling,” IEEE Trans. Signal Processing, vol. 53, no. 2, Feb. 2005.
[4] A.C. Popescu and H. Farid, “Exposing digital forgeries in color filter array interpolated images,” IEEE Trans. Signal Processing, vol. 53, no. 10, Oct. 2005.
[5] A. Swaminathan, M. Wu, and K.J.R. Liu, “Non-intrusive component forensics of visual sensors using output images,” IEEE Trans. Information Forensics and Security, vol. 2, no. 1, Mar. 2007.
[6] A. Swaminathan, M. Wu, and K.J.R. Liu, “Non-intrusive forensic analysis of visual sensors using output images,” Proc. of IEEE Conference on Acoustic, Speech and Signal Processing, 2006.
[7] A. Swaminathan, M. Wu, and K.J.R. Liu, “Component forensics of digital cameras: a non-intrusive approach,” Proc. of Conference on Information Sciences and Systems, 2006.
[8] H. Farid and S. Lyu, “Higher-order wavelet statistics and their application to digital forensics,” IEEE Workshop on Statistical Analysis in Computer Vision, 2003.
[9] S. Lyu and H. Farid, “How realistic is photorealistic?” IEEE Trans. Signal Processing, vol. 53, no. 2, Feb. 2005.
[10] M. Kharrazi, H.T. Sencar, and N. Memon, “Blind source camera identification,” Proc. of International Conference on Image Processing, 2004.
[11] M.J. Tsai and G.H. Wu, “Using image features to identify camera sources,” Proc. of IEEE Conference on Acoustic, Speech and Signal Processing, 2006.
[12] J. Lukas, J. Fridrich, and M. Goljan, “Digital “bullet scratches” for images,” Proc. of International Conference on Image Processing, 2005.
[13] J. Lukas, J. Fridrich, and M. Goljan, “Determining digital image origin using sensor imperfections,” Proc. of the SPIE, image and video communications and processing, 2005.
[14] J. Lukas, J. Fridrich, and M. Goljan, “Digital camera identification from sensor pattern noise,” IEEE Trans. Information Forensics and Security, vol. 1, no. 2, Jun. 2006.
[15] M.K. Johnson and H. Farid, “Exposing digital forgeries by detecting inconsistencies in lighting,” 7th ACM Multimedia and Security Workshop, 2005.
[16] C.C. Chang and C.J. Lin, “LIBSVM: A library for support vector machines,” 2001. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm.
[17] N.X. Lian, L. Chang, V. Zagorodnov, and Y.P. Tan, “Reversing Demosaicking and compression in color filter array image processing: performance analysis and modeling,” IEEE Trans. Image Processing, vol. 15, no. 11, Nov. 2006.
[18] L. Zhang and X. Wu, “Color Demosaicking via directional linear minimum mean square-error estimation,” IEEE Trans. Image Processing, vol. 14, no. 12, Dec. 2005.
[19] D. Alleysson, S. Süsstrunk, and J. Hérault, “Linear Demosaicing Inspired by the Human Visual System,” IEEE Trans. Image Processing, vol. 14, no. 4, Apr. 2005.
[20]R.O. Duda, P.E. Hart, and D.G. Stork, Pattern Classification, 2nd Edition, John Wiley, 2001.