在此文中，我們發展出一套使用單眼PAN-TILT攝影機的即時人臉追蹤系統。系統包含人臉偵測、可變形樣板追蹤、與運動控制。在人臉偵測上，我們提出採基因演算法搜尋人臉特徵的方法，並使用Adaboost進行人臉偵測器的訓練。在人臉追蹤上，我們提出偵測與追蹤合併的追蹤方法，藉由不斷的更新人臉樣板，達到比單一樣板追蹤更強健的追蹤法。在PAN-TILT攝影機控制上，分別採用兩個模糊控制器來追蹤移動中的人臉。最後在我們的測試下，系統可以在個人電腦上達以每秒30張的速度在複雜環境下執行人臉追蹤任務。

In this research we develop a Real-time face tracking system by single pan-tilt camera. The system includes face detection, deformable template tracking and motion control. We refer a method to search the facial features by using the genetic algorithm searching technique, the learning algorithm for face detector is based on AdaBoost. In the face tracking, we refer a tracking way to combine with detection and tracking. In the pan-tilt camera control part, two fuzzy logic controllers are designed to control the tracking and handling of moving face. We achieve a more robust tracking way than the single-template by renewing face-template continuously. Finally in our tests, the system can track the face of people in 30-frame per second under complex environment by using the personal computer.

中文摘要 I
英文摘要 II
致謝 III
目錄 IV
表索引 IX
符號索引 X
符號索引 X
第一章 概論 1
1.1研究動機 1
1.2文獻回顧 2
1.3本文架構 5
第二章 人臉偵測 6
2.1影像積分 6
2.2矩形特徵與物件偵測 9
2.3基因演算法簡介 11
2.3利用基因演算法搜尋矩形特徵 13
2.4目標影像的辨識與人臉偵測器訓練 16
2.4目標影像的辨識與人臉偵測器訓練 17
第三章 可變形樣板追蹤 22
3.1影像幾何變形(GEOMETRIC DISTORTION) 22
3.2影像的動態追蹤法 23
3.2.1 Lucas-Kanade Algorithm 23
3.2.2 Inverse Compositional Algorithm (IC) 25
3.2.3 Normalization Inverse Compositional Algorithm(NIC) 28
3.3影像區塊追蹤 30
3.4討論與比較 32
3.5 可變形樣板追蹤於人臉追蹤之應用 33
3.5.1 單一樣板追蹤 33
3.5.2 即時更新樣板追蹤 34
第四章 系統架構與控制 37
4.1 系統軟硬體設備 37
4.2 PAN-TILT CAMERA控制介紹 40
4.3 控制器設計 42
4.4系統追蹤流程 46
第五章 實驗結果 48
5.1人臉偵測訓練樣本 48
5.2人臉偵測實驗結果 48
5.3人臉追蹤實驗結果 51
5.4即時人臉追蹤實驗 53
第六章 結論與建議 56
參考文獻 59

[1] Arsenioa, A., and Santos-Victor, J.,“Robust Visual Tracking by an Active Observer”,IEEE IROS 97,Vol 3,pp1342-1347,1997.
[2] Baker, S., Gross, R., Matthews, I., and Ishikawa, T., “Lucas-Kanade 20 Years On: A Unifying Framework: Part 2”,Robotics Institute, Carnegie Mellon University, CMU-RI-TR-03-01,2003.
[3] Baker,S., and Matthews,I.,“Lucas-Kanade 20 Years On: A Unifying Framework”. International Journal of Computer Vision, 53(3), pp221–255, 2004.
[4] Baker, S., Matthews,I., Xiao,J., Gross,R., Kanade,T., and Ishikawa,T.,“Real-Time Non-Rigid Driver Head Tracking for Driver Mental State Estimation”, 11th World Congress on Intelligent Transportation Systems, October 2004.
[5] Baumberg, A.M., and Hogg, D.C.,“An Efficient Method for Contour Tracking using Active Shape Models”Motion of Non-Rigid and Articulated Objects,Proceedings of the IEEE Workshop ,pp194-199, Nov. 1994.
[6] Freund, Y., and Schapire, R.,“A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting”Journal of Computer and System Sciences, 55(1), pp119–139, Aug 1997.
[7] Hager,G.D., and Belhumeur, P.N.,“Efficient Region Tracking with Parametric Models of Geometry and Illumination” Volume 20, Issue 10, pp1025-1039 Oct. 1998.
[8] Horn, B.K.P., and Schunck,Jr.E.J.,“Determining Optical Flow” Artificial Intelligence, Vol 17, pp185-203, 1981.
[9] Jing, X., Baker, S., Matthews, I., and Kanade, T.,“Computer Vision and Pattern Recognition”,IEEE CVPR ,Vol. 2, pp535-542, July 2004.
[10] Lee,Y.B., You, B.J., and Lee, S.W.,“A Real-Time Color-Based Object Tracking Robust to Irregular Illumination Variations” IEEE ICRA. Vol.2, pp1659-1664, 2001.
[11] Lerdsudwichai,C., and Abdel-Mottaleb,M., “Algorithm for Multiple Faces Tracking”,Multimedia and Expo, ICME '03, International Conference
Vol2, pp777-780 July 2003.
[12] Lienhart, R., Liang,L. and Kuranov,A.,“A Detector Tree of Boosted Classifiers for Real-time Object Detection and Tracking”IEEE ICME 2003, Vol. 2, pp. 277-280, July 2003.
[13] Lienhart Rainer, and Maydt Jochen,“An Extended Set of Haar-like Features for Rapid Object Detection” IEEE ICIP 2002, Vol. 1, pp. 900-903, 2002.
[14] Li, S.Z., Zhang,Q., Shum, H., and Zhang, H.J.,“FloatBoost Learning for Classification” Neural Information Processing Systems 2002.
[15] Lucas,B., and Kande,T.,“An Iterative Images Registration Technique with Application to Stereo Vision”Proc DARPA Image Understanding Workshop pp121-130, 1981.
[16] Mohan, A., Papageorgiou,C., and Poggio,T.,“Example-based Object Detection in Images by Components” Pattern Analysis and Machine Intelligence, IEEE, Vol.20 ,pp39-51, 1998.
[17] Murray, D., and Basu,A.,“Motion Tracking with an Active Camera”,IEEE, Pattern Analysis and Machine Intelligence,Vol 16, pp449-459, 1994.
[18] Papageorgiou, C.P, Oren, M., and Poggio, T.,“A General Framework for Object Detection”In International Conference on Computer Vision ,pp555-562 Jan. 1998.
[19] Rowley, H.A., Baluja, S., and Kanade, T.,“Neural Network-Based Face Detection”IEEE CVPR , pp 203-208, June 1996.
[20] Sun,Y.B. Kim, J.T., and Lee,W.H.,“Extraction of Face Objects Using Skin Color Information”,Communications, Circuits and Systems and West Sino Expositions,Vol 2,pp1136-1140,2002.
[21] Viola Paul, and J.Jones. Michael“Rapid Object Detection Using a Boosted Cascade of Simple Features” IEEE CVPR ,Vol. 1, NO. 2, pp 511-518, Dec. 2001.
[22] Wu,J., Rehg, J.M., Mullin,M.D., “Learning a Rare Event Detection Cascade by Direct Feature Selection” NIPS, 2003.
[23] Zhang, Z.Q., Zhu, L., Li, S.Z., and Zhang,H.J.,“Real-time Multi-View Face Detection”Automatic Face and Gesture Recognition Proceedings. Fifth IEEE International Conference pp140-147, 2002.
[24] 邱文鴻 ,“自動人頭偵測與追蹤系統：使用一般化Hough轉換”國立成功大學資訊工程學系碩士論文, 2004.
[25] 蘇木春, 章孝德, ”機器學習:類神經網路、模糊系統、以及基因演算法則”,全華科技圖書 2002.
[26] Microsoft DirectX 9.0 SDK.
[27] Microsoft Visual C++ 6.0 Processor Pack.