人臉辨識在現今是一種非常成熟的非接觸式辨識技術，隨著電腦視覺技術飛速成長，臉部辨識在現今科技已成熱門的技術，從手機解鎖到高科技研究場所人員的身分認證都能有此技術的蹤跡，但基於RGB彩色影像的人臉辨識會受到環境或光線不佳導致識別效能不佳，也會因有心人偽裝面部騙取機器信任，進而導致重要資料外洩等情形。近年來，熱影像技術的蓬勃發展，在許多研究中也將熱影像技術應用於人臉識別，並結合RGB彩色影像可以有效的提升識別效能，熱影像的出現彌補了RGB彩色影像偽造及光照的缺點，在熱影像的生物特徵信息中，我們也可以獲得彩色影像沒有的特徵，與RGB彩色影像互補，進而提升識別效能，也能改善場地限制的問題。
本論文提出了一種新穎的特徵提取方法，將熱影像和RGB彩色影像轉換為特徵影像，並使用深度學習建構人臉辨識模型，提出的特徵萃取方法是將影像(熱影像和RGB)切割成12塊後再將其分為24和48塊作為特徵影像，再將其影像進行人臉辨識和人臉攻擊測試，實驗結果顯示， RGB影像與熱影像都能準確的辨識出人臉，而在人臉攻擊的部分，實驗證明，熱影像比RGB影像更能抵抗人臉偽裝攻擊。

Face recognition is a very mature non-contact recognition technology today. With the rapid growth of computer vision technology, facial recognition has become a popular technology in today’s technology, from unlocking mobile phones to the identification of personnel in high-tech research sites. There are traces of this technology, but face recognition based on RGB color images will be affected by the environment or poor light, resulting in poor recognition performance. It will also deceive the trust of the machine by disguising the face, which will lead to the leakage of important data. In recent years, thermal imaging technology has developed vigorously. In many studies, thermal imaging technology has also been applied to face recognition, and combined with RGB color images can effectively improve the recognition performance. The emergence of thermal images compensates for the forgery and illumination of RGB color images. Disadvantages, in the biometric information of the thermal image, we can also obtain the features that the color image does not have, which complements the RGB color image, thereby enhancing the recognition performance and improving the problem of site restrictions.
This paper proposes a novel feature extraction method that converts thermal images and RGB color images into feature images, and uses deep learning to construct a face recognition model. The proposed feature extraction method is to cut the image (thermal image and RGB) into 12 After the block, it is divided into 24 and 48 blocks as feature images, and then the images are subjected to face recognition and face attack tests. The experimental results show that both RGB images and thermal images can accurately identify human faces. In the face attack part, experiments have shown that thermal images are more resistant to facial camouflage attacks than RGB images.

中文學位論文考試審定書 ii
摘 要 iii
Abstract v
誌謝 vii
表目錄 x
圖目錄 xi
緒論 1
1.1 前言 1
1.2 研究動機與背景 5
1.3 研究方法 6
1.4 論文架構 7
文獻探討與技術分析 8
2.1紅外熱影像原理 8
2.1.1熱影像原理 9
2.2基於熱影像目標檢測 11
2.2.1臉部黃金比例 13
2.3熱影像技術探討 14
2.4人臉攻擊探討 18
研究架構與方法 21
3.1擬議系統架構 21
3.2數據收集 22
3.3影像預處理 24
3.3.1影像攝取 25
3.3.2臉部標記 26
3.3.3臉部區域分割 29
3.4特徵萃取 30
3.4.1特徵影像 33
3.5模型訓練架構 36
3.6小結 38
實驗結果與分析 39
4.1實驗目的 39
4.2實驗方法 40
4.2.1實驗環境規格 40
4.2.2人臉攻擊模擬 40
4.2.2.1照片攻擊模擬分析 41
4.2.2.2數位攻擊模擬分析 43
4.2.2.3面具攻擊模擬分析 45
4.2.2.4化妝攻擊模擬分析 47
4.3實驗數據分析 49
4.3.1人臉攻擊模擬數據分析 49
4.3.2熱影像數據分析 50
結論與展望 54
5.1結論 54
5.2未來展望 55
參考文獻 56
附件一 論文審查意見 62
附件二 論文比對結果 64

[1]中商情報網，“2021年全球生物識別行業市場現狀及發展趨勢預測分析”, [Online]. Available：https://www.163.com/ dy/article. [Accessed 2021].
[2]陸本正，2007，“基於分步降維與支援向量雞隻虹膜辨識系統”，碩士倫文，台灣科技大學。
[3]葉乘銘，2020，“應用指紋辨識於對稱式加密機制之研究”，碩士倫文，國防大學。
[4]Alex Graves, Abdel-rahman Mohamed, Geoffrey Hinton, 2013,“Speech recognition with deep recurrent neural networks”, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.
[5]羅正漢，“人臉辨識應用全面爆發！AI加持與終端應用是關鍵”，[Online]. Available：https://www.ithome.com.tw/tech/121690. [Accessed 2021].
[6]王承恩，2009，“臉型辨識技術專利分析與技術預測”，碩士倫文，中華大學。
[7]Gaurav Goswami, Mayank Vatsa, and Richa Singh,“RGB-D Face Recognition With Texture and Attribute Features”IEEE Transactions on Information Forensics and Security, pp. 1629-1640, 2014.
[8]Krishna Rao Kakkirala, Srinivasa Rao Chalamala, and Santosh Kumar Jami, “Thermal Infrared Face Recognition: A Review”2017 UKSim-AMSS 19th International Conference on Computer Modelling & Simulation (UKSim), pp. 2473-3520, 2018.
[9]Reza Shoja Ghiass, Ognjen Arandjelović, Hakim Bendada, Xavier Maldague, “Infrared face recognition: A literature review”2013 International Joint Conference on Neural Networks (IJCNN), 2013.
[10]G. Fanelli, J. Gall and L. Van Gool,“Real time head pose estimation with random regression forests”IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, 2011.
[11]F. A. Kondori, S. Yousefi, H. Li, S. Sonning and S. Sonning,“3D head pose estimation using the Kinect”2011 International Conference on Wireless Communications and Signal Processing (WCSP), Nanjing, pp. 1-4, 2011.
[12]F. Al-Khalidi, R. Saatchi, H. Elphick and D. Burke,“An evaluation of thermal imaging based respiration rate monitoring in children”American Journal of Engineering and Applied Sciences, vol. 4, no. 4, pp. 586–597, 2011.
[13]S.Ioannou, V. Gallese and A. Merla,“Thermal infrared imaging in psychophysiology: Potentialities and limits”Psychophysiology, vol. 51, no. 10, pp. 951–963, 2014.
[14]B. B. Lahiri, S. Bagavathiappan, T. Jayakumar and J. Philip,“Medical applications of infrared thermography: A review”Infrared Physics & Technology, vol. 55, no. 4, pp. 221–235, 2012.
[15]A. Rogalski,K.Chrzanowski,“Infrared devices and techniques (revision) ”Metrology and Measurement Systems., vol. 21, no. 4, pp. 565–618, 2014.
[16]I. Pavlidis, J.Levine, P.Baukol,“Thermal imaging for anxiety detection”Proceedings IEEE Workshop on Computer Vision Beyond the Visible Spectrum: Methods and Applications (Cat. No.PR00640), Hilton Head, 2002.
[17]L. B. Wolff, D. A. Socolinsky and C. K. Eveland,“Quantitative measurement of illumination invariance for face recognition using thermal infrared imagery”Proceedings of the SPIE, vol. 4820, no. 15, pp. 140-151, 2003.
[18]F. Boughorbel, B. Abidi, A. Koschan and M. Abidi,“Registration of infra-red and color images for multimodal face recognition”Biometric Symp.—Special Session Research at the Biometric Consortium Conf, Crystal City, 2004.
[19]J. Wang and E. Sung,“Facial Feature Extraction in an Infrared Image by Proxy With a Visible Face Image”IEEE Transactions on Instrumentation and Measurement, vol. 56, no. 5, pp. 2057-2066, 2007.
[20]H. Genno, A. Saijo, H. Yoshida, R. Suzuki and M. Osumi,“Using facial skin temperature to objectively evaluate sensations”International Journal of Industrial Ergonomics, vol. 19, no. 2, pp. 161–171, 1997.
[21]M. Hess and G. Martinez,“Facial feature extraction based on the smallest univalue segment assimilating nucles (SUSAN) algorithm”Picture Coding Symposium, 2004.
[22]J.-G. Wang and T. H. Lye,“Visualizing skin temperature before, during and after exercise for dynamic area telethermometry”Proceedings of the 23rd Annual International Conference of the IEEE, pp. 2831–2835, 2001.
[23]J.-G. Wang and E. Sung,“Frontal-view face detection and facial feature extraction using color and morphological operations”Pattern Recognit. Lett., vol. 20, no. 10, pp. 1053–1068, Oct. 1999.
[24]S. M. Smith and J. M. Brady,“SUSAN-A new approach to low level image processing”International Journal of Computer Vision, vol. 23, no. 1, pp. 45–78, 1997.
[25]黃葉飛揚,“PhotoShop液化濾鏡使用方法：人物面部液化塑形詳解”, [Online].Available：http://www.86ps.com/Article/PSRX/8860_P2.html. [Accessed 2021].
[26]U. Jain, B. Tan and Q. Li,“Concealed knowledge identification using facial thermal imaging”2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1677-1680, 2012.
[27]D. Li, C. C. Menassa and V. R. Kamat,“Non-intrusive interpretation of human thermal comfort through analysis of facial infrared thermography”Energy Buildings, vol. 176, pp. 246-261, 2018.
[28]R. M. A. Kiekens, A. M. Kuijpers-Jagtman, M. A. van ’t Hof, B. E. van ’t Hof and J. C. Maltha,“Putative golden proportions as predictors of facial esthetics in adolescents”Amer. J. Orthodontics Dentofacial Orthopedics, vol. 134, no. 4, pp. 480-483, Oct. 2008.
[29]M. Peter Prendergast,“Facial -proportions”Advanced Surgical Facial Rejuvenation, pp. 15-22, 2011.
[30]B. Cao, N. Wang, J. Li and X. Gao,“Data augmentation-based joint learning for heterogeneous face recognition”IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 6, pp. 1731-1743, 2019.
[31]Z. Wang, Z. Chen and F. Wu,“Thermal to visible facial image translation using generative adversarial networks”IEEE Signal Process. Lett., vol. 25, no. 8, pp. 1161-1165, 2018.
[32]S. Wang, B. Pan, H. Chen and Q. Ji,“Thermal augmented expression recognition”IEEE Transactions on Cybernetics, vol. 48, no. 7, pp. 2203-2214, 2018.
[33]C. Peng, X. Gao, N. Wang and J. Li,“Graphical representation for heterogeneous face recognition”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 2, pp. 301-312, 2017.
[34]江思賢，2018，“基於熱影像特徵之多人臉辨識系統”，碩士倫文，臺北科技大學。
[35]N. Dalal and B. Triggs,“Histograms of oriented gradients for human detection”2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), pp. 886-893, 2005.
[36]殷其佑，2020，“基於深度學習之人臉識別技術的熱特徵驗證研究”，碩士倫文，高雄科技大學。
[37]Dlib C++ library。[online].Available:http://dlib.net/ [Accessed 2021].
[38]C. Szegedy, S. Ioffe and V. Vanhoucke, 2017,“Inception-v4: InceptionResNet and the Impact of Residual Connections on Learning”31st AAAI Conference on Artificial Intelligence, AAAI 2017, pp. 4278-4284.
[39]Z. Akhtar, C. Micheloni and G. L. Foresti,“Biometric liveness detection: Challenges and research opportunities”IEEE Security Privacy, vol. 13, no. 5, pp. 63-72, 2015.
[40]鄭博仁，2019，“結合CNN和人臉熱影像之非接觸式數位身份驗證”碩士倫文，高雄科技大學。
[41]未來城市，“想躲過 AI 監控？試試俄國工程師發明的「反臉部識別」特殊化妝”[Online]. Available：https://futurecity.cw.com.tw/article/806. [Accessed 2021].
[42]M. Krišto,M. Ivasic-Kos,“An overview of thermal face recognition methods” 2018 41st International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), pp. 1098-1103, 2018.
[43]Zheng Wu, Nathan Fuller, Diane Theriault,“A Thermal Infrared Video Benchmark for Visual Analysis”2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops, 2014.