數位視訊為資訊傳遞的重要媒介，在現今廣設監控攝影機的環境下，其影像內容常被做為偵查現場或法律程序上的證據。然而，數位視訊容易編修的問題引發了若干疑慮，若有人懷著惡意而對視訊進行編輯修改，可能造成影像內容的差異而影響事後的檢視判斷結果。本研究的主要目的是鑑定數位視訊片段是否曾被編輯，可能的攻擊包括畫面插入、刪減以及替換等。研究方法主要針對於被修改視訊的二次壓縮與GOP (Group of Pictures)的關係所產生的特徵，利用H.264/AVC標準中的編碼模式資訊追蹤經由編輯操作所留下的不尋常痕跡。近期深度學習領域的蓬勃發展建立了更可靠的內容辨識技術，我們選擇使用卷積神經網路分析一連串畫面中的編碼不正常跡象，再經由偵測有規律性位置的異常畫面以判斷該視訊於首次壓縮時所使用的原始GOP大小。最後，我們以GOP與異常畫面位置等資訊推斷此視訊被編輯的實際位置。在實驗中我們設計了許多不同的測試情境，同時搭配新穎的反偵測機制進行驗證，以確認所提出方法的強健性。

Digital videos are ubiquitous these days, serving as an important source or medium for information dissemination. Since many digital surveillance cameras are deployed around cities, the content of digital videos often provides effective visual evidence of crime scene investigation or proof on the court. Nevertheless, easiness of content manipulation raises certain concerns over authenticity of digital videos. A malicious user may tamper the video via widely available editing tools to change the meaning of content so that the subsequent examination or analysis would be affected. This research aims at providing a video forensic tool for determining whether an investigated video has been edited. The considered editing operations include frame/video segment insertion or deletion. The methodology is to employ the fact that video editing usually results in double compression/encoding and affects the coding selection of certain frames when the fixed GOP (Group of Pictures) is used. The coding modes of H.264/AVC are utilized to examine the traces of video editing via convolutional neural networks. The abnormal frames appearing periodically will be located to determine the original GOP size of the first encoding, which helps to identify the exact frame or video editing location. Several testing cases are designed in the experiments, coupled with the state-of-the-art anti-forensic approach, to verify the feasibility of the proposed method.

[1] K. Sitara, B.M. Mehtre, “Digital video tampering detection: An overview of passive techniques.” Digital Investigation, vol.18, pp. 8-22, 2016.
[2] Y. Tew, K.S. Wong, “An overview of information hiding in H.264/AVC compressed video,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 24, no. 2, pp. 305-319, 2014.
[3] R.D. Singh, N. Aggarwal, “Video content authentication techniques: a comprehensive survey,” Multimedia Systems, vol. 24, pp. 211–240, 2018
[4] X. Jiang, W. Wang, T. Sun, Y. Q. Shi and S. Wang, “Detection of Double Compression in MPEG-4 Videos Based on Markov Statistics,” IEEE Signal Processing Letters, vol. 20, no. 5, pp. 447-450, 2013.
[5] W. Wang, H. Farid, “Exposing digital forgeries in video by detecting double MPEG compression,” Proceedings of the 8th Workshop on Multimedia and Security, pp. 37–47, 2006
[6] T. Wiegand, G.J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 13, Issue. 7, pp. 560–576, 2003.
[7] C.-K. Hsieh, “Detecting Shot Manipulation in H.264/AVC Videos by Analyzing the Changes of Macroblock Coding Modes,” National Central University, Master’s Thesis, 2018
[8] W. Wang and H. Farid, “Exposing digital forgeries in video by detecting double quantization,” Proceedings of the 11th ACM workshop on multimedia and security, pp. 39–48, 2009.
[9] A. Gironi, M. Fontani, T. Bianchi, A. Piva, M. Barni, ”A video forensic technique for detecting frame deletion and insertion,” IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 6226–6230, 2014.
[10] Y. Su, J. Xu, “Detection of double-compression in MPEG-2 videos,” 2010 2nd International Workshop on Intelligent Systems and Applications, pp. 1–4, 2010.
[11] P.-C. Su, P.-L. Suei, M.-K. Chang, J. Lain, “Forensic and anti-forensic techniques for video shot editing in H.264/AVC,” Journal of Visual Communication and Image Representation, vol.29, pp. 103–113, 2015.
[12] T. Sun, W. Wang, X. Jiang, “Exposing video forgeries by detecting MPEG double compression,” IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1389–1392, 2012.
[13] J. Xu, Y. Su, X. You, “Detection of video transcoding for digital forensics,” IEEE International Conference on Audio, Language and Image Processing, pp. 160–164, 2012.
[14] P. Bestagini, A. Allam, S. Milani, M. Tagliasacchi, S. Tubaro, “Video codec identification,” IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 2257–2260, 2012.
[15] G. Yammine, E. Wige, A. Kaup, “Blind GOP structure analysis of MPEG-2 and H.264/AVC decoded video,” 28th Picture Coding Symposium, pp. 258–261, 2010.
[16] W. Wang, X. Jiang, S. Wang, M. Wan, T. Sun, “Identifying Video Forgery Process Using Optical Flow,” Revised Selected Papers Digital-forensics and watermarking: 12th International workshop, pp. 244–257, 2014.
[17] J. Chao, X. Jiang, T. Sun, “A novel video inter-frame forgery model detection scheme based on optical flow consistency,” Revised Selected Papers Digital forensics and watermarking: 11th International workshop, pp. 267–281, 2013.
[18] L. Zheng, T. Sun, Y.-Q. Shi, “Inter-frame video forgery detection based on block-wise brightness variance descriptor,” Revised Selected Papers Digital-forensics and watermarking: 13th International Workshop, pp.18-30, 2015.
[19] M.C. Stamm, W.S. Lin, K. Liu, “Temporal forensics and anti-forensics for motion compensated video,” IEEE Transactions on Information Forensics and Security, vol.07, no.04, pp. 1315–1329, 2012.
[20] Q. Wang, Z. Li, Z. Zhang, Q. Ma, “Video inter-frame forgery identification based on consistency of correlation coefficients of gray values,” Journal of Computer and Communications, vol.02, no.04, pp. 51-57, 2014.
[21] Y. Su, J. Zhang, J. Liu, “Exposing digital video forgery by detecting motion-compensated edge artifact,” 2009 International Conference on Computational Intelligence and Software Engineering, pp. 1–4, 2009.
[22] H. Liu, S. Li, S. Bian, “Detecting frame deletion in H.264 video,” Information Security Practice and Experience Conference, pp.267-270, 2014.
[23] D. Vazquez-Padin, M. Fontani, T. Bianchi, P. Comesaña, A. Piva, M. Barni, “Detection of video double encoding with GOP size estimation,” IEEE International Workshop on Information Forensics and Security, pp.151-156, 2012.
[24] B. Sahiner et al, “Classification of mass and normal breast tissue: a convolution neural network classifier with spatial domain and texture images,” in IEEE Transactions on Medical Imaging, vol.15, no.5, pp.598-610, 1996.
[25] S. Ioffe, C. Szegedy, “Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift,” in ICML, 2015.
[26] YUV Dataset. [Online]. Available: https://media.xiph.org/video/derf/.
[27] JM.18 Reference Software. [Online]. Available: http://iphome.hhi.de/suehring/tml/download/.
[28] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, I. Polosukhin, "Attention Is All You Need," In Neural Information Processing Systems, 2017