中國相學是一門經過長時間統計與傳承的學問，含括了面相學、手相學、痣相學與骨相學等之探討。面相學由於直觀性較高，在民間常運用於性格的判斷與運勢的推估；而在中國傳統醫學中所使用“望聞問切”中的「望診」，更是透過觀察面相，評估氣色，來診斷患者的健康狀況。現今科技發達，個人資料往往相當容易暴露於網路上。社群網站、電信公司以及金融機構，如管理不善，都有洩漏個人資料的疑慮。另一方面，近年來詐騙集團猖獗，他們會利用非法取得的個人資料，大量偽造身份，騙取財物，危害金融與社會秩序。警政署為了防範民眾受騙，開始宣導防詐騙手法以及設置防詐騙專線。不過，詐騙手法日新月異，每年台灣地區詐騙金額接近十億元新台幣，使得國民人心惶惶。因此，吾人希望能藉由觀察人臉中的五官，針對其全域性以及區域性特徵，做分析與評量，設計出一套可靠與完善的人臉辨識系統，以加強個人資料之控管。
本論文探討人臉偵測與辨識系統之設計與實作策略。人臉偵測部分，吾人利用色彩空間轉換與影像分群技術，將人臉與背景分離；其次，運用人眼偵測技術，找出確切的人臉位置；最後，運用局部二元圖像(Local Binary Patterns)，分別針對人臉五官，以全域性及區域性特徵萃取的方法，建立人臉模型，做合法使用者之判定。本系統在CPU時脈為2.3GHz的Intel Core i7個人電腦與Windows 7作業系統環境下，以獨立式數位相機作為輸入裝置時，對實驗室與民眾，共16人，每人拍攝10張照片，使用全域性與區域性二元圖像，正確辨識率可分別達到98.61%與100%；以筆電內建相機作為輸入裝置時，對實驗室10人，每人拍攝10張照片，使用全域性與區域性二元圖像，正確辨識率可分別達到100%與100% 。而透過PUT、FEI以及HK PolyU所提供的人臉資料庫，分別各有100、200以及335人，系統正確辨識率亦可分別達到100%、96.00%以及83.28%。

Chinese chirology is derived from long term observation and everlasting inheritance, comprising physiognomy on face, palm, mole and bone. Facial physiognomy is often used to determine character and foretell divination of a person due to its explicit features. Four examinations, visual inspection, listening and smelling, inquiry, and palpation, are jointly used in traditional Chinese medicine to evaluate the health condition of a person. Visual inspection is first applied to collect the clues of illness through observing the facial complexion and color of a patient. The advancement of information technology makes personal data much more easily exposed to the internet than before. Community websites, telecommunication companies and financial institutions are all the potential sources for improper disclosure of personal data if adequate security measurement is not adopted. In the recent years, fraudulent activities have been prevailing in our society, a significant amount of fake IDs are created, people are kept deceived, and public orders are endangered. Although anti-fraudulence movement have been advocated by National Police Agency to mitigate the effects of disorder, the amount of loss is still climbed to near one billion dollars a year due to the fast-changing fraudulent tricks. People are frightened, and effective measures must be taken. Therefore, it is our objective to analyze the properties of the global and regional features of the face, and design a practical and reliable face recognition system to greatly reduce the risk of ID faking.
This thesis investigates the design and implementation strategy of a face detection and recognition system. For the face detection, color space transformation and clustering techniques are first applied to separate the human skin from the background. Then, the central coordinates of eyes are found to locate the candidate region of the face. Finally, local binary patterns are used to extract both the global and regional texture features of faces, build probability models for each individual and determine the legitimate user. Two image capturing devices, a digital camera and a notebook webcam, are chosen to acquire the facial pictures in this study. Under the 2.3GHz Intel Core i7 PC and Windows7 operating system environment, a correct face recognition rates of 98.61% and 100% can be reached respectively using the global and regional features for a digital camera database of 16 persons with 10 pictures each person. The correct rates of 100% and 100% can also be achieved respectively using the global and regional features for a notebook webcam database of 10 persons with 10 pictures each person. In addition, the system is tested on PUT, FEI and HK PolyU NIR Face Databases with 100, 200 and 335 users respectively, the recognition rates of 100%、96.00% and 83.28% can be accomplished respectively.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
目錄 v
圖次 vii
表次 x
第一章 緒論 1
1.1研究動機 1
1.2研究方法 3
1.3論文章節概要 3
第二章 面相學 4
2.1人臉簡介 4
2.2三停 5
2.3十二宮 6
2.4眉毛 8
2.5眼睛 10
2.6鼻子 13
2.7嘴巴 15
2.8耳朵 17
第三章 相關文獻探討 21
3.1色彩空間 21
3.1.1 NCC色彩空間 21
3.1.2 YCbCr色彩空間 23
3.2人臉辨識 24
3.2.1卡式轉換 24
3.2.2局部二元圖像 26
第四章 影像辨識系統之流程與架構 28
4.1影像辨識流程簡介 28
4.2影像前處理 29
4.2.1膚色偵測 30
4.2.2人臉邊界 33
4.2.3人眼偵測 35
4.2.4去除臉部反光 36
4.2.5標記法 38
4.2.6人眼座標 45
4.2.7人臉切割 46
4.3人臉特徵萃取 47
4.4影像模型比對 51
第五章 辨識系統之實作成果與效能評析 52
5.1軟硬體設備與開發平台 52
5.2人臉模型建立與訓練 53
5.3 權重分析 55
5.4 系統辨識效能 58
第六章 結論與未來展望 67
參考文獻 68

[1] 內政部警政署全球資訊網： http://www.npa.gov.tw/NPAGip/wSite/mp?mp=1
[2] 面相圖解： http://tiger168.com/hanface/168fac00.htm
[3] 維基百科：http://zh.wikipedia.org/
[4] 鐘國亮，影像處理與電腦視覺(第五版)，台北市；東華書局，民國103年。
[5] A. McAndrew, J.-H. Wang and C.-S Tseng，數位影像處理（劉震昌譯），臺北市；高立圖書，民國100年。
[6] 門協尚平，觀人術(謝肇琳、曾稚 合譯)，台北市；五洲，民國86年。
[7] 齊祖詮，人相簡述(施存、施敏 出版)，新竹市；三億打字商行，民國94年。
[8] 周湘靈，開運面相大不同，台北市；小知堂文化事業有限公司，民國93年。
[9] 飛雲山人，看相從頭學，台北市；時報文化出版企業有限公司，民國81年。
[10] João Ascenso, Paulo Correia and Fernando Pereira, “A face detection solution integrating automatic and user assisted tools,” 11th Portuguese Conference on Pattern Recognition, Porto, Portugal, pp.1-8, 2000.
[11] Rein-Lien Hsu, Mohamed Abdel-Mottaleb and Anil K. Jain, “Face detection in color images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 24, No. 5, pp. 696-706, 2002.
[12] Chih-Chien Thomas Chen, Chin-Ta Chen and Ming-Hong Jiang, “Hard-limited Karhunen-Loeve transform for face recognition,” IEICE Trans. Fundamentals of Electronics, Communications and Computer Sciences, Vol. E87-A, No. 7, pp. 1836-1838, 2004.
[13] Timo Ojala, Matti Pietikäinen and Topi Mäenpää, “Multiresolution gray-scale and rotation invariant texture classification with local binary patterns,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, No. 7, pp. 971-987, 2002.
[14] Vu Thi Lan Huong, Dong-Chul Park, Dong-Min Woo and Yunsik Lee, “Centroid neural network with Chi square distance measure for texture classification,” Proceedings of International Joint Conference on Neural Networks, pp. 1310-1315, 2009.
[15] Timo Ahonen, Abdenour Hadid and Matti Pietikäinen, “Face description with local binary patterns: Application to face recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28, No. 12, pp. 2037-2041, 2006.
[16] Zhenhua Guo, Lei Zhang, David Zhang and Xuanqin Mou, “Hierarchical multiscale LBP for face and palmprint recognition,” International Conference on Image Processing, IEEE - ICIP, pp. 4521-4524, 2010.
[17] Chin-Chuan Han, Hong-Yuan Mark Liao, Gwo-Jong Yu and Liang-Hua Chen, “Fast face detection via morphology-based preprocessing,” Pattern Recognition, Vol. 33, No. 10, pp. 1701-1712, 2000.
[18] Xiao-Yi Feng, “Eyes location by neural network-based face segmentation,” IJCSNS International Journal of Computer Science and Network Security, Vol. 6, No. 5A, pp. 132-135, 2006.
[19] Maricor Soriano, Birgitta Martinkauppi and Sami Huovinen, “Skin detection in video under changing illumination conditions,” 15th International Conference on Pattern Recognition, IEEE – ICPR, Vol. 1, pp. 839-842, 2000.
[20] Kwok-Wai Wong, Kin-Man Lam and Wan-Chi Siu, “A robust scheme for live detection of human faces in color images,” Signal Processing-image Communication, Vol. 18, No. 2, pp. 103-114, 2003.
[21] Christophe Garcia and George Tziritas, “Face detection using quantized skin color regions merging and Wavelet packet analysis,” IEEE Transactions on Multimedia – TMM, Vol. 1, No. 3, pp. 264-277, 1999.
[22] H. C. Vijay Lakshmi and S. Patil Kulakarni, “Segmentation algorithm for multiple face detection in color images with skin tone regions using color spaces and edge detection techniques,” International Journal of Computer Theory and Engineering, Vol. 2, No. 4, pp. 552-558, 2010.
[23] CIE Biometrics https : //biometrics.cie.put.poznan.pl/
[24] A. Kasiński, A. Florek and A. Schmidt “THE PUT FACE DATABASE,” Image Processing & Communications, Vol. 13, No. 3-4, pp. 59-64, 2008.
[25] FEI Face Database : http://fei.edu.br/~cet/facedatabase.html
[26] The Hong Kong Polytechnic University (PolyU) NIR Face Database : http://www4.comp.polyu.edu.hk/~biometrics/polyudb_face.htm
[27] C. E. Thomaz and G. A. Giraldi, “A new ranking method for principal components analysis and its application to face image analysis,” Image and Vision Computing, Vol. 28, No. 6, pp. 902-913, 2010.
[28] V. Amaral and C. E. Thomaz, “Normalizacao espacial de imagens frontais de face,” Technical Report 01/2008 (in portuguese), Department of Electrical Engineering, FEI, São Bernardo do Campo, São Paulo, Brazil, 2008.
[29] Baochang Zhang, Lei Zhang, David Zhang and Linlin Shen, “Directional binary code with application to PolyU near-infrared face database,” Pattern Recognition Letters, Vol. 31, issue 14, pp. 2337-2344, 2010.