臺灣博碩士論文加值系統

　　目前人臉辨識系統之研究大致可區分成兩類。第一類是人臉表情之分析，從人臉五官中擷取其特徵點，利用這些特徵點的相對關係以分析人臉的表情。第二類是人臉身分的辨識，取出人臉的獨特特徵，以辨認其身分。本論文之研究屬於後者，以辨識人臉的身分為目標，從輸入具有完整臉孔的彩色影像，到最後的人臉身分識別，建構出一套自動化臉孔辨識系統。本系統內容包括人臉偵測、臉部特徵定位、臉部特徵擷取以及臉孔辨識四個主要模組。
　　人臉偵測模組，採用YCbCr色彩模型，藉以切割出人臉與背景；臉部特徵定位模組，先利用瞳孔初步定位眉毛、眼睛、鼻子、嘴巴的相對位置，再經由臉部特徵擷取模組精確定位以取得眼睛、鼻子、嘴巴的特徵點及眉毛形狀的特徵值；臉孔辨識模組，以混合式類神經網路為架構，其輸入為特徵值及特徵點所構成的四組局部特徵向量與一組整體特徵向量，藉此辨識出影像中人臉的身分。實驗資料使用兩組影像，一組是自拍的影像，另一組是CMU PIE(CMU Pose, Illumination, and Expression, PIE)影像資料庫。測試結果中，自拍影像之平均正確辨識率高達95%，而PIE資料庫之平均正確辨識率為83%。若將辨識結果的前兩名列入考量，則自拍影像與PIE影像的辨識率分別可達100%與91.2%。由此可知，本研究所提出之自動化臉孔辨識系統具有極佳的辨識率與可靠度。

　　The face recognition systems can be roughly divided into two categories. One is the analysis of face expression, which first extracts facial feature points and then uses their relative relation to analyze the face expression. The other is the identification of a to-be-recognized face, which extracts the unique character of face and then identifies the face. This research, belonging to the latter, is to build an automatic face recognition system that accepts a color image with a complete face and identifies the face. The system consists of four modules: face detection module, facial features location module, facial features extraction module and face recognition module.
　　The face detection module adopts the YCbCr color model to separate the face and the background. The facial features location module uses the pupils to locate the relative positions of eyebrows, eyes, nose and mouth roughly. After that, the facial features extraction module exactly locates the positions the facial features, extracts the pre-defined feature points and calculates the moments of the shapes of the eyebrows in order to obtain four local feature vectors and one global feature vector. Finally, the face recognition module uses a hybrid structure, consisting of a RBF and a multilayer neural network, to identify the to-be-recognized face. The experiments are tested by two sets of image database. One database, including 28 persons and 20 images per person, is photoed by us and the other is the well-known partial CMU PIE image database, including 68 persons and 50 images per person. The average recognition accuracy of our images is 95% and that of PIE is 83%. If we put the first two candidates of recognized result into consideration, the accuracies achieve 100% and 91.2% respectively. From the experimental results, the proposed automatic face recognition system is of effective recognition rate and of high reliability.

中文摘要……………………………………………………………………………I
Abstract……………………………………………………………………………II
目錄………………………………………………………………………………IV
表目錄……………………………………………………………………………VI
圖目錄……………………………………………………………………………VII

第一章 緒論……………………………………………………………………… 1
1.1 研究動機與目的……………………………………………………… 1
1.2 相關研究..……………………………….……………………………… 2
1.3 論文架構….…………………………………………………………… 3

第二章 系統架構流程…………………………………………………………… 6

第三章 人臉偵測與臉部特徵定位……………………………………………… 10
3.1 人臉偵測…..…………………………………………………………… 10
3.1.1 膚色色彩分析…...…………………………………………………… 11
3.1.2 臉部瞳孔分析…...…………………………………………………… 15
3.1.3 人臉區塊切割…...…………………………………………………… 19
3.2 臉部特徵定位…..……………………………………………………… 21

第四章 臉部特徵擷取........................................................................................... 23
4,1 眉毛特徵擷取….……………………………………………………… 24
4.1.1 眉毛形狀擷取…..…………………………………………………… 24
4.1.2 眉毛形狀特徵與特徵點…..………………………………………… 29
4.2 眼睛特徵擷取….……………………………………………………… 32
4.3 鼻子特徵擷取………………………………………………………… 36
4.4 嘴巴特徵擷取………………………………………………………… 39
4.5 整體特徵……………………………………………………………… 47

第五章 臉孔辨識.................................................................................................... 49
5.1 特徵值計算….………………………………………………………… 49
5.2 類神經網路架構….…………………………………………………… 50

第六章 實驗結果………………………………………………………………… 51
6.1 測試影像資料………………………………………………………… 51
6.2 實驗結果……………………………………………………………… 54
6.2.1 自拍影像資料庫.…………………………………………………… 54
6.2.2 PIE影像資料庫...…………………………………………………… 57

第七章 結論…………………………………………………………………… 67

參考文獻……………………………………………………………………… 68

附錄A 系統操作流程……………………………………………………………72

附錄B 肖像權合約

[3] Andrew J. Calder, A. Mike Burton, Paul Miller, Andrew W. Young, Shigeru Akamatsu, “A principal component analysis of facial expressions,” Vision Research, Volume: 41, Issue: 9, pp. 1179 – 1208, April, 2001

[4] Christophe Garcia, Manolis Delakis, “A neural architecture for fast and robust face detection,” Proceedings of 16th International Conference on Pattern Recognition, Volume: 2, 11-15, pp.44-47, Aug. 2002

[5] Douglas Chai, Abdesselam Bouzerdoum, ”A Bayeslan Approach to Skin Color Classification in YCBCR Color Space,” Proceeding of TENCON 2000, Vol.2, pp.421-424, 2000.

[6] G..C. Feng, P.C. Yuen, “Variance projection function and its application to eye detection for human face recognition,” Pattern Recognition Letters, Volume: 19, Issue: 9, pp. 899-906, July, 1998

[7] Hongxun Yao, Wen Gao, “Face detection and location based on skin chrominance and lip chrominance transformation from color images,” Pattern Recognition, Volume: 34, Issue: 8, pp. 1555-1564, August, 2001

[8] Javad Haddadnia, Majid Ahmadi, “N-feature neural network human face recognition,” Image and Vision Computing, Volume: 22, Issue: 12, pp. 1071-1082, October 1, 2004

[9] Jen-Tzung Chien, Chia-Chen Wu, “Discriminant Waveletfaces and Nearest Feature Classifiers for Face Recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.24, no.12, December, 2002

[10] Jianxin Wu, Zhi-Hua Zhou, “Efficient face candidates selector for face detection,” Pattern Recognition, Volume: 36, Issue: 5, pp. 1175-1186, May, 2003

[11] Juan J. de Dios, Narciso Garcia, “Face detection based on a new color space YCgCr,” 2003 International Conference on Image Processing, Volume: 3, Pages:III 909-912 vol.2, Sept.14-17, 2003

[12] Kyung-Yung Choi, Kunio Takaya, “Facial feature extraction from a video sequence using independent component analysis,” 2001 IEEE Pacific Rim Conference on Computers and signal Processing, Volume 1, pp.259-262, 26-28 Aug, 2001

[13] L. Jordao, M. Perrone, J.P. Costeira, J. Santos-Victor, “Active face and feature tracking,” Proceedings of International Conference on Image Analysis and Processing, pp.572-576, 1999

[14] Ming-Kuei Hu, “Visual Pattern Recognition by Moment Invariants,” IRE Trans. Info. Theory, Vol. IT-8, pp. 179-187

[15] Ming-Hsuan Yang, Narendra Ahuja, David Kraegman, “Face recognition using kernel eigenfaces,“ 2000 International Conference on Image Processing, Volume: 1, Pages:37-40 vol.1, 10-13 Sept. 2000

[16] Paola Campadelli, Raffaella Lanzarotti, “Fiducial point localization in color images of face foregrounds,” Image and Vision Computing, Volume: 22, Issue: 11, pp. 863-872, September 20, 2004

[17] R. L. Hsu, M. Abdel-Mottaleb, A. K. Jain, “Face detection in color images,” IEEE Transaction on Pattern Analysis and Machine Intelligence, 24(2), 2002, pp. 696-706

[18] S. Birchfield, “Elliptical head tracking using intensity gradients and color histograms,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.232-237, 1998

[19] Shu-Hung Leung, Shi-Lin Wang, Wing-Hong Lau, ”Lip Image Segmentation Using Fuzzy Clustering Incorporating an Elliptic Shape Function,” IEEE Transactions on Image Processing, Vol. 13, No. 1, January, 2004

[20] Simon Lucey, Sridha Sridharan, Vinod Chandran, “Adaptive mouth segmentation using chromatic features,” Pattern Recognition Letters, Volume: 23, Issue: 11, pp. 1293-1302, September, 2002

[21] Tien C. Hsia, senior member, IEEE, “A Note on Invariant Moments in Image Processing,” IEEE Transactions on System Man and Cybernetics, Vol. SMC-11, no. 12, December 1981.

[22] W.S. Lee, H. J. Lee, J. H. Chung, “Wavelet-based FLD for face recognition,” Proceedings of the 43rd IEEE Midwest Symposium on Circuits and Systems, Volume: 2, 8-11 Aug. 2000. Pages: 734-737 vol.2

[23] Zhenhe Chen, Xiao-Ping Zhang, “Video sequence processing based on spatiotemporal independent component analysis,” IEEE CCECE 2003 Canadian Conference on Electrical and Computer Engineering, Volume 2, pp.1207-1210 vol.2, 4-7 May 2003

[24] Zhi-Hua Zhou, Xin Geng, “Projection functions for eye detection,” Pattern Recognition, Volume: 37, Issue: 5, May, 2004, pp. 1049-1056

[25] 吳明衛，自動化臉部表情分析系統，碩士論文，國立成奶j學資訊工程系，2003，6月。

[26] 吳瑞珍，人臉特徵自動抽取之演算法設計與應用，碩士論文，元智大學電機工程研究所，2001，6月。

[27] 陳文賢，應用小波技術於複雜背景之人臉偵測，碩士論文，義守大學電子工程學系，2003，7月。

[28] 陳彥良，即時立體視覺物體追蹤系統，碩士論文，中原大學機械工程學系，2003，7月。

[29] Rafael C. Gonzales, Richard E. Woods, “Digital Image Processing, 2/E,” Prentice Hall International Editions, pp.672-675 ISBN: 0-13-09465-8

[30] Sergios Theodoridis, Konstantinos Koutroumbas, “Pattern Recognition,” Academic Press, pp.245-248, ISBN: 0-12-686140-4

[31] Simon Baker, “The CMU Pose, Illumination, and Expression (PIE) database,” http://www.ri.cmu.edu/projects/project_418.html