The application of digital image processing has been spread widely and constantly because of the improvements in the speed and efficiency of computers in recent years. Every new human-machine inter-face not only solves many problems of people’s lives but also plays an important role in the control field [1-2]. Some researchers are so interested in the problem that they replace traditional hand controlling by audio and video signals [3]. Grattan, K T [4-5] developed a microcomputer-based system for disabled simply rely upon differential reflection from the eyeball. The input blink could be used to activate the presentation of the data as a binary tree or matrix scan. With eyeball tracking, any change of the eyes has become a new command [6]. But image-based eyeball tracking system with auto-zooming CCD camera will give more freedom to users. Besides, the hardware used in image-based eyeball tracking system can cooperate with videoconferencing equipment. An eye-tracking interface determine the position of a user’s pupil and maps this position into a point of the display screen [7]. This is a great convenience in life. Usually the device includes a detecting device to detect bio-electromagnetic signals or image sequence generated by eye movements. To some handicapped, the image-based eyeball tracking technology seems more convenient and outstanding. That is our research motivation.

A major part of the eyeball tracking technology is the feature extraction of the eyes. There have been some reports about it in recent years. The research work can be devided into two forms, namely (a) the one based on the synthesis of local information such as edges and (b) the one based on deformable template method using global information [8]. Irrespective of the kind of approach, the search work of eyeball is important and time consuming. Since the eyes are a small part of the face, the research is in order to find out the approximate location of the eye window, then extract and analyze it.

index