臺灣博碩士論文加值系統

　　隨著電腦視覺技術的發展，基於視覺方式的偵測與辨識以提供人類與電腦 (或智慧型手機) 間非接觸式的互動技術，已成為日常生活應用的重要研究課題。
　　本研究的主旨在開發一套「具學習能力之客製化臉部表情辨識系統」。傳統臉部表情辨識大多是根據一組含有許多使用者之大型資料庫進行訓練，本系統採取不同的方式，主要是設計來提供一種可客製化的方法，可以使用非負矩陣分解法學習並建立樣板臉部表情 (即無表情、高興、生氣、驚訝、傷心)。本系統可分為兩部分說明：(1) 臉部表情辨識方法；(2) 臉部表情辨識操作流程。為了進行系統評估，10位使用者參與本研究，研究結果顯示系統可初步成功的辨識五種特定表情。
　　總結而言，本研究明顯證明本系統可提供臉部表情辨識的潛在解決方案，同時可以詮釋臉部表情因人而異的現象。

Through the development of computer vision technology, vision-based detection &; recognition capable of providing contactless interaction between human and computers (or smart phones) have become research of interest for daily applications.
In this study, our objective is to develop a “customizable facial expression recognition system with learning capability”. Different from conventional facial expression recognition that is often trained using a large database with multiple users, our system is designed to provide a customizable approach that can learn and establish template facial expression (i.e., normal, happy, angry, surprise, and sad) using the Non-Negative Matrix Factorization for further recognition. The system can be described in two parts: (1) facial expression recognition method, and (2) facial expression recognition operation flow. For system evaluation, 10 users participated in this study and our results demonstrate preliminary success for recognizing the five specific facial expression.
In summary, this study clearly shows that our system may provide a potential solution to facial expression recognition, addressing the inter-user variation in facial expression.

摘要 ... I
Abstract ... II
致謝 ... III
目錄 ... IV
圖索引 ... VI
表索引 ... VIII
第一章　緒論 ... 1
1.1 研究背景 ... 1
1.2 相關研究 ... 2
1.3 研究目的 ... 3
1.4 論文架構 ... 4
第二章　基礎理論 ... 5
2.1 彩色影像轉灰階 ... 5
2.2 應用AdaBoost之人臉偵測 ... 6
2.2.1 AdaBoost演算法 ... 6
2.2.2 Haar特徵與積分影像 ... 7
2.2.3 層疊分類器(Cascade Classifier) ... 10
2.3 以基底為主之非負矩陣分解法 ... 11
2.4 餘弦值 ... 14
第三章　研究方法 ... 15
3.1 臉部表情辨識方法 ... 15
3.1.1 臉部偵測 ... 17
3.1.2 樣板臉部表情與未知臉部表情 ... 17
3.1.3 強調基底之非負矩陣分解法 ... 18
3.1.4 降維 ... 19
3.1.5 相似度 ... 20
3.2 系統操作流程 ... 20
第四章　研究結果 ... 24
4.1 研究環境 ... 24
4.2 客製化樣板表情之基底影像 ... 25
4.3 相似度計算結果 ... 29
4.4 辨識結果 ... 30
4.5 表情辨識誤判分析 ... 34
第五章　討論與未來展望 ... 35
參考文獻 ... 36

圖2-1 彩色影像轉灰階圖 ... 6
圖2-2 AdaBoost演算法之分類器架構 ... 7
圖2-3 Haar矩形特徵 ... 8
圖2-4 A(x,y) 位置之積分影像值 ... 8
圖2-5 某矩形特徵區塊之灰階值總和 ... 10
圖2-6 層疊分類器示意圖 ... 10
圖3-1 客製化臉部表情辨識系統方法之方塊圖 ... 16
圖3-2 臉部偵測之影像示意圖 ... 17
圖3-3 灰階影像矩陣放置之方法 ... 18
圖3-4 降維示意圖 ... 19
圖3-5 客製化臉部表情辨識系統操作流程圖 ... 21
圖3-6 系統在樣板階段之操作流程 ... 22
圖3-7 系統在辨識階段之操作流程 ... 23
圖4-1 系統之鏡頭前拍攝條件 ... 25
圖4-2 迭代次數 T=30 之臉部基底影像 ... 26
圖4-3 迭代次數 T=80 之臉部基底影像 ... 26
圖4-4 迭代次數 T=150 之臉部基底影像 ... 26
圖4-5 客製化樣板表情與基底影像資料庫－範例一 ... 27
圖4-6 客製化樣板表情與基底影像資料庫－範例二 ... 28
圖4-7 客製化樣板表情與基底影像資料庫－範例三 ... 28
圖4-8 相似度計算示意圖 ... 29
圖4-9 辨識結果－使用者一 ... 31
圖4-10 辨識結果－使用者二 ... 32
圖4-11 系統之表情辨識誤判分析 ... 34

表4-1 系統研究使用之硬體設備 ... 24
表4-2為本系統實驗之總結果 ... 33

［1］M. A. Turk and A. P. Pentland, “Face Recognition Using Eigenfaces,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 586-591, Jun 1991.
［2］Y. Xue, C. S. Tong and W. Zhang, “Evaluation of Distance Measures for NMF-based Face Recognition,” International Conference on Computational Intelligence and Security, vol. 1, pp. 651-656, Nov 2006.
［3］T. Zhang, B. Fang, Y. Y. Tang, G. He and J. Wen, “Topology Preserving Non-negative Matrix Factorization for Face Recognition,” IEEE Trans. Image Processing, vol. 17, pp. 574-584, Apr 2008.
［4］S. G. Kong, J. Heo, B. R. Abidi, J. Paik, M. A. Abidi, “Recent Advances in Visual and Infrared Face Recognition—A Review,” Computer Vision and Image Understanding, vol. 97, no. 1, pp. 103-135, Jan. 2005.
［5］J. Calder, A. M. Burton, P. Miller, A. W. Young and S. Akamatsu, “A Principal Component Analysis of Facial Expressions,” Vision Research, vol. 41, no. 9, pp. 1179-1208, Apr 2001.
［6］I. Buciu and I. Pitas, “Application of Non-negative and Local Non Negative Matrix Factorization to Facial Expression Recognition,” 17th International Conference on Pattern Recognition, vol. 1, pp. 288-291, Aug 2004.
［7］H. Sako, A. V. W. Smith, “Real-time Facial Expression Recognition Based on Features' Positions and Dimensions,” 13th International Conference on Pattern Recognition, vol. 3, pp. 643-648, Aug 1996.
［8］Yu Zhang, E. C. Prakash, E. Sung, “Real-time Physically-based Facial Expression Animation Using Mass-spring System,” Proceedings of Computer Graphics International, pp. 347-350, 2001.
［9］P. Michel, R. E. Kaliouby, “Real Time Facial Expression Recognition in Video Using Support Vector Machines,” Proceedings of the 5th international conference, on Multimodal interfaces, pp. 258-264, 2003.
［10］Y. Wang, H. Ai, B. Wu, C. Huang, “Real Time Facial Expression Recognition with AdaBoost,” 17th International Conference on Pattern Recognition, ICPR, vol. 3, pp. 926-929, Aug 2004.
［11］P. Viola, M. Jones, “Robust Real-time Face Detection,” IEEE International Conference on Computer Vision, ICCV, Proceedings, vol. 2, pp. 747, 2001.
［12］Y. Freund, R. Schapire, “A Decision-theoretic Generalization of On-line Learning and an Application to Boosting,” Journal of Computer and System Sciences, pp. 119 - 139, 1997. 55(1):119–139, 1997.
［13］D. D. Lee and H. S. Seung, “Algorithms for non-negative matrix factorization,” Adv. Neural Inf. Process, pp. 556 - 562, 2000.
［14］P. P. Ou, “Face Recognition Using Basis-emphasized Non-negative Matrix Factorization with Wavelet Transform,” Master Thesis, Dept. Computer Science &; Information Engineeing, National Taiwan University, 2006(In Chinese).