隨著科技日新月異的進步，視覺科技產品也越來越多樣，如3D電視的研發及具有3D互動效果之遊樂器等產品。因為其新穎的功能而成功的吸引使用者去購買其產品，但其產品因為往往須配戴其專屬的眼鏡，長時間使用會增加使用者眼睛的負擔。因此，此論文利用影像處理提出了一套有效的即時人眼追蹤之方法，根據使用者不同的角度觀看螢幕，並且配合建構之3D情境，讓使用者達到類似觀看3D電視之效果。在此論文中，此系統主要分為四大步驟：先基於膚色偵測及邊緣偵測等方法得到所需的資訊，由所得之資訊偵測出人臉範圍，再依據人臉之範圍，找出人眼的所在位置，最後再根據人眼與螢幕之相對位置旋轉螢幕內之3D情境。在人臉偵測方面使用了影像的垂直及水平直方圖來進行偵測，人眼偵測方面使用重心方法來進行人眼偵測，3D情境部分則使用OpenGL建構。由實驗結果可知，本論文所研究之人眼追蹤方式，可以在使用時容忍外在之干擾，以增加系統之穩定性，此外，本論文使用較低之影像解析度，減少系統計算時間，使系統更為流暢。

Along with improvements in technology, products that bridge human-machine interactions are becoming more common, like 3D television and game consoles which have 3D function, etc. These types of products are popular because of their innovative features. However, the particular glasses must be worn when using these products, and may affect eye comfort with extended use. Therefore, this thesis developed an image-process-based system that tracks human eyes, the 3D model of which is built according to the relative position of the eyes and the screen. The proposed system includes four parts: information about skin region detection and edge detection; face region detection according to previous information; eye position detection in face regions; and rotation of the 3D model by calculation of the relative position between the eyes and the screen to rotate the 3D model. For face detection, the system detects the maximum edge regions in vertical and horizontal histograms of the image; while for eyes tracking, it detects eye position by using the centroid method. Further, the 3D model uses OpenGL to draw 3D models. The simulation result shows that this eye-tracking algorithm can resist external disturbances. Moreover, this thesis uses lower DPI (dots per inch) to reduce the computing time, making it more efficient.

[1] T.Y. Chen, Design and implementation of the face tracking and recognition system, Department of Engineering Science National Cheng Kung University, 2007.
[2] F. Gasparini and R. Schettini, “Skin segmentation using multiple thresholding,” Proceedings of SPIE, vol. 6061, pp. 1-8, 2006.
[3] M. Hu, S. Worrall, A. H. Sadka and A. M. Kondoz, “Face feature detection and model design for 2-D scalable model-based video coding”, Visual Information Engineering, pp. 125-128, 2003
[4] M.C. Jing, A simple image processing demo under .Net environment using C#, 2007.
[5] K. Karadimitriou and J. M. Tyler, “The centroid method for compressing sets of similar images,” Pattern Recognition Letters, vol. 19, pp. 585-593, 1998.
[6] S. Kherchaoui and A. Houacine, “Face detection based on a model of the skin color with constraints and template matching,” International Conference on Machine and Web Intelligence, pp. 469-472, 2010.
[7] H. Li, X. Wu, D. Zhao and J. Zhai, “An algorithm of skin detection based on texture,” International Congress on Image and Signal Processing, vol.4, pp.1822-1825, 2011.
[8] M. Segal and K. Akeley, “The design of the OpenGL graphics interface.” Silicon Graphics technical report, 1994.
[9] M. Segal and K. Akeley, The OpenGL graphics system: a specification (Version 1.2.1). Silicon Graphics Computer Systems, 1998.
[10] M. Soriano, B. Martinkauppi, S. Huovinen, and M. Laaksonen, “Skin detection in video under changing illumination conditions,” Proceedings of International Conference on Pattern Recognition, vol.1, pp. 839-842, 2000.
[11] S. Thakur. “Face detection using skin tone segmentation,” Information and Communication Technologies, pp. 53-60, 2011.
[12] R.S. Wright, B. Lipchak and N. Haemel, OpenGL superbible, 4th ed, Addison-Wesley, Upper Saddle River, NJ, 2007.
[13] G. Woods, Digital image processing, 3/e, Pearson Education Taiwan Ltd, 2009.