7.1 結論
我們提出了一個多視角的人臉追蹤方法，利用粒子濾波做為主要的追蹤方 式。其中我們修改了轉移性函式，在追蹤的過程中能根據人臉移動的情況動態的 改變粒子散佈的範圍，讓追蹤能更具強韌性(robust)；相似性函式則是利用動態檢 查表做為目標比對的依據，可以根據目前的勢姿找出動態檢查表裡對應的人臉資 訊來比對進而達到多視角的人臉追蹤。
為了能在不同人臉角度的變化下持續進行追蹤，我們使用了多視角的人臉判 斷機制來取得目前影格中目標區域的人臉姿勢，針對不同姿勢的歷史記錄儲存到 動態檢查表之中，同時提供了更新機制可以隨著追蹤的過程動態反應目前人臉的 變化，讓動態檢查表能保存較新、較接近目前人臉狀態的歷史資訊來達到更準確 的追蹤。
針對我們的追蹤方法我們也提出了校正機制，對於追踨時人臉突然消失的狀 況做即時的處理，藉由改變取樣的方式加上動態檢查表的比對而能有效地搜尋影 像裡的目標，並在找到目標的同時重新設定追踨所需的參數，這樣的設計可以使 發生追踨錯誤的情況能大幅度的降低，讓整體系統更具可靠性(reliability)與容錯 性(tolerable)。
7.2 未來的研究工作
對於我們提出的追蹤方法，我們認為下面幾點是在未來可以繼續延伸的內 容：
(1) 目前我們只針對左右側臉的旋轉進行角度的切割，對於上下點頭的角度 變化是利用人臉判斷時加入上下點頭的人臉訓練資料當作判斷的容忍 度。如果將人臉非平面旋轉角度進行適當的分割並建立了對應的動態檢 查表，如圖 22 所示，就像是建立了一個在不同角度下簡化的人臉流形 曲面(manifold)，對於更多樣的角度變化可以有更精準的判斷。不過動態 檢查表要在不同的姿勢間切換或移動也會相對變得複雜，如何有效率地 進行不同姿勢的比對又不失其精確性，是一個值得研究的議題。
TR
θt
TR
xt TR
yt TR
wt TR
ht
圖 23. 人臉流形曲面示意圖。其中的灰色部 分是我們目前動態檢查表處理的範圍。
圖24. 修改後的目標區域表示法。
(2) 我們的追蹤方法無法處理太大的平面旋轉角度，主要的原因在於粒子濾 波追蹤機制所使用的樣本並沒有考慮平面旋轉的情況。其改進的方法可
)
以將目標區域與樣本表示法多加一個表示平面旋轉的角度，以目標區域 為例，可以表示成TRt =(xTRt ,yTRt ,wtTR,htTR,θtTR ，如圖23 所示，便可以更 精確的表示真實人臉的狀態，不過相對而言，轉移性函式就必須多考慮 一個維度的參數變化，所使用的樣本數必須增加才能表示出目前影格的 目標區域所有可能的變化情形，這樣便會增加處理一張影格所需的時 間。這又是一個精確度與速度之間的考量。
(3) 我們所提出的人臉追蹤技術必須使用人臉判斷機制做為找尋人臉姿勢 的方式，目前我們所使用的人臉判斷是針對不同姿勢的人臉各別訓練一 組 LDA 人臉判斷器，主要是希望為動態檢查表提供一個快速簡單的判 斷功能。不過目前多視角的人臉偵測技術已經相當成熟，我們希望未來 可 以 結 合 人 臉 偵 測 所 使 用 的 技 術 ， 例 如 Haar-like features [15] 與 Adaboost-like 的判斷機制[21], [14], [24]等來達到更精確的人臉判斷。
參考文獻
[1] M. Arulampalam, S. Maskell, N. Gordon, and T. Clapp, "A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking," IEEE Transactions on Signal Processing, vol. 50, no. 2, pp. 174-188, 2002.
[2] F. Dornaika and F. Davoine, "On Appearance Based Face and Facial Action Tracking," IEEE Transactions on Circuits and Systems for Video Technology, vol.
16, no. 9, pp. 1107-1124, 2006.
[3] F. Fleuret and D. Geman, "Coarse-to-Fine Face Detection," International Journal of Computer Vision, vol. 41, no. 1-2, pp. 85-107, 2001.
[4] W. Frank, A. Markus, and R. Gerhard, "Multimodal Face Detection, Head Orientation and Eye Gaze Tracking," IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems, pp. 13-18, 2006.
[5] S. Gokturk, J. Bouguet, and R. Grzeszczuk, "A data-driven model for monocular face tracking," Proceedings of Eighth IEEE International Conference on
Computer Vision, vol. 2, pp. 701-708, 2001.
[6] H. Grabner, M. Grabner, and H. Bischof, "Real-time tracking via on-line
boosting," Proceedings of British Machine Vision Conference, pp. 47–56, 2006.
[7] C. Huang, H. Ai, Y. Li, and S. Lao, "High-Performance Rotation Invariant
Multiview Face Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 4, pp. 671-686, 2007.
[8] M. Isard and A. Blake, "CONDENSATION - Conditional Density Propagation for Visual Tracking," International Journal Computer Vision, vol. 29, no. 1, pp.
5-28, 1998.
[9] A. Jacquot, P. Sturm, and O. Ruch, "Adaptive Tracking of Non-Rigid Objects Based on Color Histograms and Automatic Parameter Selection," Proceedings of the IEEE Workshop on Motion and Video Computing, vol. 2, pp. 103-109, 2005.
[10] T.S. Jebara and A. Pentland, "Parametrized structure from motion for 3D adaptive feedback tracking of faces," Proceedings of the Conference on Computer Vision and Pattern Recognition, pp. 144-150, 1997.
[11] M. Jones and P. Viola, "Fast multi-view face detection," Mitsubishi Electric Research Lab TR-20003-96, 2003.
[12] S. Karlsson, M. Taj, and A. Cavallaro, "Detection and tracking of humans and faces," EURASIP Journal on Image and Video Processing, vol. 2008, no. 11, pp.
1-9, 2008.
[13] K. Lee, J. Ho, M. Yang, and D. Kriegman, "Visual tracking and recognition using probabilistic appearance manifolds," Computer Vision and Image Understanding, vol. 99, no. 3, pp. 303-331, 2005.
[14] S. Li and Z. Zhang, "FloatBoost learning and statistical face detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 26, no. 9, pp.
1112-1123, 2004.
[15] R. Lienhart and J. Maydt, "An extended set of Haar-like features for rapid object detection," Proceedings of International Conference on Image Processing, vol. 1, pp. I-900-I-903, 2002.
[16] P. Pérez, C. Hue, J. Vermaak, and M. Gangnet, "Color-Based Probabilistic
Tracking," Proceedings of the 7th European Conference on Computer Vision-Part I, vol. 2350/2002, pp. 661-675, 2002.
[17] R. Qian, M. Sezan, and K. Matthews, "A robust real-time face tracking
algorithm," Proceedings of International Conference on Image Processing, vol. 1, pp. 131-135, 1998.
[18] D.A. Ross, J. Lim, R. Lin, and M. Yang, "Incremental Learning for Robust Visual Tracking," International Journal of Computer Vision, vol. 77, no. 1-3, pp.
125-141, 2008.
[19] H. Stern and B. Efros, "Adaptive color space switching for face tracking in multi-colored lighting environments," Proceedings of Fifth IEEE International Conference on Automatic Face and Gesture Recognition, pp. 249-254, 2002.
[20] M. Turk and A. Pentland, "Eigenface for recognition," Journal of Cognitive Neuroscience, vol. 3, no. 1, pp. 71-86, 1991.
[21] P. Viola and M. Jones, "Robust Real-time Object Detection," International Journal of Computer Vision, vol. 57, no.2, pp.137-154, 2002.
[22] J. Wang, X. Chen, and W. Gao, "Online selecting discriminative tracking features using particle filter," IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2, pp. 1037-1042, 2005.
[23] Z. Wen and T. Huang, "Capturing subtle facial motions in 3D face tracking,"
Proceedings of Ninth IEEE International Conference on Computer Vision, pp.
1343-1350, 2003.
[24] J. Wu, S. Brubaker, M. Mullin, and J. Rehg, "Fast Asymmetric Learning for Cascade Face Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 30, no. 3, pp. 369-382, 2008.
[25] T. Yang, S.Z. Li, Q. Pan, J. Li, and C. Zhao, "Reliable and Fast Tracking of Faces under Varying Pose," Proceedings of the 7th International Conference on
Automatic Face and Gesture Recognition, pp. 421-426, 2006.
[26] T. Yang, Z. Li, Q. Pan, J. Li, C. Zhao, and Y. Cheng, "Online Adaptive Fast Multipose Face Tracking Based on Visual Cue Selection," Acta Automatica Sinica, vol. 34, no. 1, pp. 14-20, 2008.
[27] Z. Yao and H. Li, "Tracking a Detected Face with Dynamic Programming,"
Proceedings of the 2004 Conference on Computer Vision and Pattern Recognition Workshop, pp. 63, 2004.
[28] Intel, Open Source Computer Vision Library, http://opencv.willowgarage.com/wiki/.