我們利用廣為人用的PCA統計模型建構出的特徵人形模組，結合平均移動追蹤法發展出一套新的追蹤人形的方法。一般的平均移動追蹤注重在藉由利用矩形形狀的追蹤。我們的新方法注重在利用更精確的人形預測去貫穿整部影片地追蹤人形。這整篇論文著重在找出人形的位置，辨別前景物體是否為人形，用改進過的平均移動法來追蹤辨認出的人形物體。最後是改進過的平均移動追蹤法中的人形主成分係數的更新方法，用以當人形形狀改變時更新它。

A new approach to human tracking was developed through the use of a popular statistical model, known as Principal Component Analysis, for constructing the human eigen-shape model in conjunction with the mean shift tracking algorithm. The regular mean shift tracker focuses on tracking through the use of the rectangle shape. The new approach focuses on using a more precise estimation of a human shape to track the human throughout a video sequence. The entire paper focuses on the localization of human; identification of the human; tracking the identified human by the modified mean shift algorithm. Last but not least, is the updating of the human principal component coefficients in the modified mean shift tracker, so as to update the human shape as it changes.

[1] Forsyth, D.A, and Ponce, J. (2002). Computer vision: a modern approach. Prentice Hall.
[2] Stauffer, C., and Grimson, W.E.L. (1999). Adaptive background mixture models for real- time tracking. Computer Vision and Pattern Recognition, Volume 2, 23-25.
[3] Comaniciu, D., Ramesh, V., and Meer, P.(2000). Real-time tracking of non-rigid objects using mean shift. Computer Vision and Pattern Recognition, Volume 2, 13-15.
[4] Jongwoo, L., and David, K. (2004). Tracking humans using prior and
learned representations of shape and appearance. IEEE International
Conference on Automatic Face and Gesture Recognition.
[5] Fatih, P., and Oncel, T. (2003). Human body tracking by adaptive
Background models and mean-shift analysis .MERL.
[6] Jing-Ping, J., Yan-Ning, Z., Rong-Chun, Z., and Ze-Tao, J.(2005). Tracking
of Objects in image sequences using multi-freeness mean shift algorithm.
IEEE.
[7] Comaniciu, D., and Meer, P. (1999). Mean shift analysis and applications. IEEE, 1197-1203.
46
[8] Jia, J.P., and Zhao, R.C.(2004). Tracking of objects in images sequences using bandwidth matrix mean shift algorithm. International Conference on Signal Processing, VOL.2, 918-921.
[9] Jepson, A.D., Fleet, D. J., and El-Maraghi, T.F. (2001). Robust on-line appearance models for visual tracking. IEEE Conference on Computer Vision andPattern Recognition, Volume I, 415–422.
[10] Zhou, J., and Hoang, J. (2005). Real time robust human detection
and tracking system. IEEE Conference on Computer Vision and Pattern
Recognition.
[11] Cuce, I. H., and Cetin, E.A.(2004). Mean shift tracking of moving
objects using multi-dimensional histograms. Signal and Data
Processing of Small Targets.
[12] Comaniciu, D., Ramesh, V., and Meer, P. (2000). Kernel- based object tracking. Real-Time Vision and Modeling Department.
[13] Ukrainitz, Y. and Sarel, B. Mean Shift Theory and Applications. Retrieved February 3, 2007, from http://www.wisdom.weizmann.ac.il/~ deniss/vision_ spring04/files/mean_shift/mean_shift.ppt.
[14] Hariharan, R. (2005). Eigen values eigen vectors and principal component. Retrieved February 3, 2007, from http://drona.csa.iisc.ernet.in/~ramesh/ courses.html.
47
[15] Fukunaga, K. (1990). Introduction to Statistical Pattern Recognition. Retrieved January 10, 2007, from http://en.wikipedia.org/wiki/Karhunen-Lo%C3%A8ve_transform
[16] Ying, W., and Ting, W. (2006). A field model for human detection and
tracking. IEEE, Vol.28, 753-764.
[17] Blake, A. and Isard, M.(2001). Active Contours. International Conference on Image Analysis and Processing, 354.
[18] Blake, A. and Isard, M.( 1996). Contour Tracking by Stochastic Propagation of Conditional Density. Proc. European Conf. Computer Vision, 343-356.
[19] Lipton, J.A., Fujiyoshi, H. and Patil, S.R.,(1998).Moving Target
Classification and Tracking from Real Time Video. IEEE.
[20] Epanechinikov Kernel. Retrieved March 16, 2007, from
http://en.wikipedia. org/wiki/Kernel_(statistics).
[21] Video grabber Software. Retrieved November 12, 2006, from
http://www. mitov.com/
[22] Borland Builder C++. Retrieved February 12, 2007, from
http:// info.borland. com/techpubs/borlandcpp/