日前人機互動介面的使用方式越來越多樣化，其中，基於種種優點，以電腦視覺的軌跡追蹤與辨識方式，是目前研究領域的一個熱門題目。又人類能夠整日移動瞳孔方向而不會感到疲勞，因此瞳孔追蹤是一個能夠整日使用的理想人機介面互動方式。此外，在移動裝置上，尤其是筆記型電腦、平板電腦、智慧型手機，大多配備了前鏡頭，因此瞳孔追蹤系統的安裝可望在不改動使用者硬體的狀況下完成。由於以上的好處，越來越多的研究機構和消費電子公司，都積極投入瞳孔追蹤控制的研究。
然而，人的眼睛和瞳孔，在前鏡頭的占據面積並不大，而且容易受到低光源雜訊的干擾，導致偵測準確率並不高，從而在偵測與追蹤的過程之中，很容易發生誤判，從而產生錯誤定位與雜點。因此，在本篇論文中，我們提出改良過的瞳孔偵測方法，並使用卡爾曼濾波器，以及Cam-Shift追蹤法，來達成較好的追蹤結果，提升瞳孔軌跡辨識的正確率。
本篇論文題出兩種瞳孔控制方法，其中一個即是瞳孔軌跡辨識，藉由辨識到的瞳孔移動行為，來控制電腦。另一個則是使用眼睛注視的區域控制滑鼠指標位置。此兩種系統能夠投入日常使用，特別是需要非接觸式的使用者介面時，或者是提供殘疾人士使用。
本篇論文使用類神經網路進行瞳孔軌跡辨識，達到80%以上的正確率。由此實證，揭示了本篇論文的系統具有足夠的強健性提供使用者使用。

The motion tracking is currently a popular research topic by the use of computer vision. Human beings move their pupils all the time without feeling tired. Therefore, pupil tracking is an ideal input method for non-contact computer interface for daylong use. In mobile devices, such as tablets, and smart phones, the front facing camera can be utilized for pupil tracking directly. Because of the benefits of pupil tracking, more and more institutes and companies are developing this kind of technology.
Because the eyes and pupil occupies small ratio in the frames of a front facing camera, and the quality of front facing cameras is limited due to some manufacture reasons, the trajectory from the detection is rather noisy especially in low light condition. Therefore, in this thesis, we proposed our pupil detection methods and use tracking methods as well as Kalman filter to eliminate noisy data, and obtain better and acceptable recognition results.
Two human computer interface are designed. One is a pupil posture recognition system, which recognizes pupil movement trajectory as commands. The other is a pupil-mouse mapping, which detects the gaze position to move the cursor of a computer. The proposed methods and systems can be employed into daily use of humans, especially for the occasions the interface must be non-contact, and those who are mentally challenged.
The neural network classifiers are employed and tested for our proposed systems. The accuracy of the pupil posture is above 80% according to our experimental results, which means our proposed method is robust enough for daily use.

[1]“SentiGaze Eye Movement Tracking System,” [Online] Available http://www.neurotechnology.com/press_release_sentigaze.html (accessed on June 26, 2015)
[2]H. F. Ho, "Reading Process of Integrated Circuit Layout Debugging: Evidence from Eye Movements," Advanced Materials Research, vol. 787, pp. 855-860, 2013
[3]R. Ballagas, J. Borchers, M. Rohs, and J. G. Sheridan, "The smart phone: a ubiquitous input device," Pervasive Computing, vol. 5, no. 1, pp. 70-77, 2006.
[4]S. Instruments, "iView X HED," ed, 2009. [Online] Available http://www.smivision.com/en/gaze-and-eye-tracking-systems/products/overview.html (accessed on June 26, 2015)
[5]T. Yoshioka, S. Nakashima, J. Odagiri, H. Tomimori, and T. Fukui, "Pupil detection in the presence of specular reflection," in Proceedings of the Symposium on Eye Tracking Research and Applications, pp. 363-364. Florida, 2014.
[6]“The Eye Tribe,” [Online] Available https://theeyetribe.com/ (accessed on June 12, 2015)
[7]"Anadolu CV Eye Tracker," [Online] Available https://www.youtube.com/watch?v=8x0dW9BdkcU (accessed on July 3, 2015)
[8]R. Lienhart and J. Maydt, "An extended set of haar-like features for rapid object detection," in Proceedings of International Conference on Image Processing, vol. 1., pp. 900-903, New York, 2002.

[9]P. Viola and M. J. Jones, "Robust real-time face detection," International Journal of Computer Vision, vol. 57, no. 2, pp. 137-154, 2004.
[10]T. Mita, T. Kaneko, and O. Hori, "Joint haar-like features for face detection," in Proceedings of International Conference on Computer Vision, pp. 1619-1626, Beijing, China, 2005.
[11]Y. Freund, R. Schapire, and N. Abe, "A short introduction to boosting," Journal-Japanese Society for Artificial Intelligence, vol. 14, no. 5, pp. 1612, 1999.
[12]A. Kasinski and A. Schmidt, "The architecture and performance of the face and eyes detection system based on the Haar cascade classifiers," Pattern Analysis and Applications, vol. 13, no. 2, pp. 197-211, 2010.
[13]D. Comaniciu and P. Meer, "Mean shift: A robust approach toward feature space analysis," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 5, pp. 603-619, 2002.
[14]D. Exner, E. Bruns, D. Kurz, A. Grundhöfer, and O. Bimber, "Fast and robust CAMShift tracking," in Proceedings of Computer Vision and Pattern Recognition Workshops, pp. 9-16, San Francisco, CA, 2010.
[15]E. R. Dougherty, and R. A. Lotufo, Hands-on morphological image processing vol. 71, SPIE press Bellingham, 2003.
[16]N. Otsu, "A threshold selection method from gray-level histograms," Automatica, vol. 11, no. 1, pp. 23-27, 1975.
[17]T. Pavlidis, "Algorithms for shape analysis of contours and waveforms," IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. PAMI-2, no. 4, pp. 301-312, 1980.
[18]R. E. Kalman, "A new approach to linear filtering and prediction problems," Journal of Fluids Engineering, vol. 82, no. 1, pp. 35-45, 1960.
[19]D. E. Rumelhart, G. E. Hinton, and R. J. Williams, "Learning representations by back-propagating errors," Cognitive Modeling, vol. 5, no. 3, p. 213-220, 1988.
[20]A. Goh, "Back-propagation neural networks for modeling complex systems," Artificial Intelligence in Engineering, vol. 9, no. 3 , pp. 143-151, 1995.
[21]Z. Zhang and A. R. Hanson, "3D reconstruction based on homography mapping," in Proceedings of ARPA Image Understanding Workshop, pp. 1007-1012, Cambridge, MA, 1996.
[22]J. M. Henderson, "Human gaze control during real-world scene perception," Trends in cognitive sciences, vol. 7, no. 11, pp. 498-504, 2003.