人臉變形攻擊對各種以人臉識別為基準的安全機制構成了嚴重的威脅，目前也尚未有可靠的方法來對抗這類攻擊。在本論文中提出了兩種基於光響應非均勻性雜訊特性提取特徵的方法，並結合兩種特徵，對由兩個不同資料集生成的多種變形參數變形圖像進行了實驗，分析出最有效的變形偵測方法，並與其他方法進行對比，證明了該方法的優勢。另外，針對變形圖片，我們提出一種結合實時圖像，從變形圖像中提取出變形源圖像成分，由此預測變形來源的方法，並證明了該方法的有效性。

The face morphing attack posed a serious threat to various security mechanisms based on face recognition, and there is no reliable approach to deal with the attack. In this research, we proposed two kinds of feature extraction methods based on the photo response non-uniformity (PRNU) and combine these two features to carry out the experiments. The experimental morphed images are generated from two different data sets. The PRNU of a face image is detected firstly, then the feature of the celled histograms of the DFT magnitudes of the detected PRNU are created. The bin with the maximum value in the celled histogram is used as an indicator to judge the real and morphed images. These bins with maximum value extracted from each cell is aggregated to calculate the root mean square as the global score S for the image. This score is used to decide if it is a morphing image. According to the experiments, the proposed method is outperformed then other methods. In addition, we propose a method to predict source face image by extracting the original face information from morphed face image and real-time face image, which simulates the real two-images detection scenario, and prove the effectiveness of this method.

[1] N. K. Ratha, J. H. Connell, R. M. Bolle, "Enhancing security and privacy in biometrics-based authentication systems", IBM Systems Journal, vol. 40, no. 3, 2001, pp. 614-634.
[2] R. Raghavendra, C. Busch, "Presentation Attack Detection Methods for Face Recognition Systems: A Comprehensive Survey", ACM Comput. Surv., vol. 50, no. 1, 2017.
[3] R. Raghavendra, K. B. Raja, C. Busch, "Detecting morphed face images", Proc. IEEE 8th Int. Conf. Biometrics Theory Appl. Syst. (BTAS), 2016, pp. 1-7.
[4] U. Scherhag et al. , "On the vulnerability of face recognition systems towards morphed face attacks", Proc. Int. Workshop Biometrics Forensics (IWBF), 2017, pp. 1-7.
[5] Best Practice Technical Guidelines for Automated Border Control (ABC) Systems—V2.0 , Warsaw, Poland, Aug. 2012.
[6] Acuity Market Intelligence, “The Airport Automated Biometric Facilitation Report: From Curb to Gate”, 2018.
[7] International Air Transport Association (IATA), IATA Automated Border Control Map, Mar. 2019, [Online]. Available: https://www.iata.org/whatwedo/passenger/Pages/automated-border-control-maps.aspx
[8] D. E. King , "Dlib-ml: A machine learning toolkit", Journal of Machine Learning Research, 2009, vol. 10.
[9] G. Wolberg, "Digital Image Warping", IEEE computer society press, Los Alamitos, CA, 1994.
[10] M. Chen, J. Fridrich, M. Goljan, J. Lukas, "Determining image origin and integrity using sensor noise", IEEE Transactions on information forensics and security, vol. 3, no. 1, Mar. 2008, pp. 74-90.
[11] M. Goljan, J. Fridrich, and J. Lukas, “Camera identification from printed images” , Proceedings of SPIE, Jan. 2008.
[12] J. Fridrich , "Digital image forensic using sensor noise", IEEE Signal Processing Magazine, vol. 26, no. 2, Mar. 2009.
[13] PyWavelets, Wavelet Properties Browser, 2019, [online]. Available: http://wavelets.pybytes.com/wavelet/db8/.
[14] M. Ferrara, A. Franco, D. Maltoni, “The magic passport”, IEEE International Joint Conference on Biometrics, 2014, pp. 1–7.
[15] FIDELITY European Project Web Site, Nov. 2017, [online]. Available: http://www.fidelity-project.eu/.
[16] A. Makrushin, T. Neubert, J. Dittmann , “Automatic generation and detection of visually faultless facial morphs”, International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, 2017, pp. 39–50
[17] M. Hildebrandt, T. Neubert, A. Makrushin, J. Dittmann,“Benchmarking face morphing forgery detection: Application of StirTrace for impact simulation of different processing steps”, International Workshop on Biometrics and Forensics, 2017, pp. 1–6.
[18] C. Kraetzer, A. Makrushin, T. Neubert, M. Hildebrandt, J. Dittmann, “Modeling attacks on photo-id documents and applying media forensics for the detection of facial morphing”, The ACM Workshop on Information Hiding and Multimedia Security, 2017, pp. 21–32.
[19] C. Seibold, W. Samek, A. Hilsmann, P. Eisert, "Detection of face morphing attacks by deep learning", International Workshop Digit. Watermarking (IWDW), 2017, pp. 107-120.
[20] D. King, "High Quality Face Recognition with Deep Metric Learning", 2017, [Online]. Available: http://blog.dlib.net/2017/02/high-quality-face-recognition-with-deep.html
[21] Merge Face API (V1), Face++, 2019, [Online]. Available: https://console.faceplusplus.com.cn/documents/20813963.
[22] A. Martínez, R. Benavente, "The AR face database", 1998.
[23] Psychological Image Collection at Stirling (PICS), 2D face sets, 2008, [Online]. Available: http://pics.psych.stir.ac.uk/2D_face_sets.htm.
[24] Information Technology—Biometric Data Interchange Formats—Part 5: Face Image Data, 2012.
[25] D. J. Robertson et al. , "Fraudulent ID using face morphs: Experiments on human and automatic recognition" in PloS One, 12(3), e0173319, 2017.
[26] T. Neubert et al. , "Extended StirTrace Benchmarking of Biometric and Forensic Qualities of Morphed Face Images" in IET Biometrics, 2018.
[27] P. Viola, M. J. Jones, "Robust real-time face detection", International Journal of Computer Vision, vol. 57, no. 2, 2014, pp. 137-154.
[28] Information Technology—Biometric Presentation Attack Detection—Part 3: Testing and Reporting, 2017.