臺灣博碩士論文加值系統

頭部姿勢是一種肢體語言，常作為人與人溝通的方式之一，因此研究良好的頭部動作辨識系統，能幫助人機介面更佳人性化。在本文提出一個使用主軸分析(Principal Component Analysis, PCA)的頭部姿勢辨識系統，先以主軸分析得到各角度人臉的特徵臉(Eigenface)，之後借助高斯混合模型(Gaussian Mixture Model, GMM)有良好近似各類分布的特性，能更佳地對各角度的特徵臉進行分群，並利用隱藏式馬可夫模型(Hidden Markov Model, HMM)能學習時間軸資訊的特性，作為辨識在時間軸上經由人臉角度變化所組成的點頭與搖頭動作。本論文的測試資料庫共有50個片段，其中影片內容包含了有白平衡上的變化、多個頻率不同的頭部連續動作、人臉遮蔽與輕微人臉表情變化。本論文所提出的方法能對該測試資料庫達到極高的辨識率。

Head gesture is one kind of body language which is often used to communicate among people. Developing a good head gesture recognition system will let the user interface more humanistic. In this paper, we propose a head gesture recognition system with principal component analysis (PCA). At first, we use PCA to analyze multi-angle head gesture images to obtain the eigenface of each face image. Then we use the Gaussian mixture model (GMM) to cluster high-dimensional feature vector. Finally we use the hidden Markov model (HMM) for the recognition of head nodding and shaking. The HMM could learn the temporal information that discriminates various face angles. The testing database includes 50 video segments with challenging characteristics such as white balance, non-uniform head movements, occlusion, and variations of face expressions. The proposed method has high recognition rate verified through the testing database.

摘要 i
英文摘要 ii
誌謝 iii
目　錄 iv
表目錄 vi
圖目錄 vii
第一章 緖論 1
1.1 研究背景 1
1.2 相關研究 2
第二章 系統架構 7
2.1 影像強化 7
2.2 特徵萃取 9
2.3參數調整與實驗分析 10
第三章 特徵萃取 11
3.1主軸分析 12
3.2 特徵分析 15
第四章 辨識系統 19
4.1 隱藏式馬可夫模型 20
4.2高斯混合模型與隱藏式馬可夫模型之結合 26
4.2高斯混合模型 27
4.2.1高斯混合模型函數參數估測法 28
第五章 實驗結果 31
5.1 實驗環境 31
5.2 影像強化 32
5.2.1 色彩模型轉換 32
5.2.2 特徵圖 34
5.3系統參數與實驗分組 38
5.4 實驗結果 40
5.5 實驗結果之個案分析 47
5.5.1辨識正確之個案分析 47
5.5.2錯誤個案分析 56
第六章 結論 57
參考文獻 59

[1] V. Kruger and G. Sommer, “Gabor wavelet networks for efficint head pose estimation,” Image and Vision Computing, vol. 20, pp. 665-672, Mar. 2002.
[2] L. Cerrato and M. Skhiri, “A method for the analysis and measurement of communicative head movements in human dialogues,” In Proceedings of Auditory-Visual Speech Proceeding, pp. 251-256, 2003.
[3] J. Sherrah and S. Gong ‘‘Fusion of perceptual cues for robust tracking of head pose and position,’’ Pattern Recognition, vol. 34, pp. 1565-1572, 2001.
[4] J. Sherrah, S. Gong, and E.J. Gong, ‘‘Face distribution in similarity space under varying head pose,’’ Image and Vision Computing, vol. 19, pp. 807-819, 2001.
[5] W. Tan and G. Rong, ‘‘A real time head nod and shake detector using HMM,’’ Expect System with Application. vol. 25, pp. 461-466, 2003.
[6] P. Lu, M. Zhang, X. Zhu, Y. Wang, “Head nod and shake recognition based on multi-view model and hidden markov model,” In Proceedings of the Computer Graphics, Imaging and Vision, 2005.
[7] A. Kapoor and R. W. Picard, “A real-time head nod and shake detector,” Workshop on Perceptive User Interfaces, Nov. 2001.
[8] H. I. Choi and P. K. Rhee, ‘‘Head gesture recognition using HMMs,’’ Expert System with Application, vol. 17, pp. 213-221, 1999.
[9] S. Kawato and J. Ohya, ‘‘Real-time detection of nodding and head-shaking by directly detecting and tracking the between eyes,’’ IEEE International Conference on Automatic Face and gesture recognition, pp. 26-30, Mar. 2000.
[10] M. W. Lee and S. Ranganath, ‘‘3D deformable face model for pose determination and face synthesis image analysis and processing,’’ 10th International Conference on Image Analysis and Processing, pp. 260-265, 1999.
[11] M. Turk and A. Pentland, “Face Recognition Using Eigenface,” In proceedings IEEE Conference on Computer Vision and Pattern Recognition, pp. 586-591, 1991.
[12] A. M. Martinez and A. C. Kak, ‘‘PCA versus LDA,’’ IEEE Transactions on Pattern Analysis and Machine Intellgence, vol. 23, no. 2, Feb. 2001.
[13] Y. S. Ryu and S. Y. Oh, “Automatic extraction of eye and mouth field from a face image using eigenfeature and multilayer perceptions,” Pattern Recognition, vol. 34, pp. 2459-2466, 2001.
[14] X. Liu, T. Chen, and S. M. Thornton, “Eigenspace updating for non-stationary process and it’s application to face recognition,” Pattern Recognition, vol. 36, pp. 1945-1959, 2003.
[15] H. A. Rowley, S. Baluja, and T. Kanade, ‘‘Neural network-based face detection,’’ IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20, pp. 23-38, Jan. 1998.
[16] J. Huang, X. Shao, and H. Wechsler, ‘‘Face pose discrimination using support vector machines (SVM),’’ In Proceedings of 14th International Conference on Pattern Recognition, pp. 155-156, 2001.
[17] J. Yu General, ‘‘C-means clustering model,’’ IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 27, pp. 1197-1211, Aug. 2005.
[18] R. Lawrence Rabiner, ‘‘A tutorial on hidded markov model and selected application in speech recognition,’’ Processing of the IEEE, vol. 77, pp. 257-286, Feb. 1989.
[19] L. E Baum and T. Petrie, ‘‘Statistical inference for probabilistic function of finite state markov,’’ Chains Annals of Math. Statistics, vol. 37, no. 1, pp. 554-1563, 1966.
[20] X. Li, M. Parizeau, and R. Plamondon, ‘‘Training hidden markov model with multiple observations-a combinatorial method,’’ IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, pp. 371-377, Apr. 2000.
[21] D. A. Ross and R. S. Zemel, “Multiple cause vector quantization,” In Advances in Neural Information Processing Systems, vol. 15, pp.1-8, 2003.