Date of Award


Document Type

Thesis - ECU Access Only


Edith Cowan University

Degree Name

Bachelor of Science Honours


School of Computer and Information Science


Faculty of Computing, Health and Science

First Supervisor

Michael Collins

Second Supervisor

Paul Patak


Humans and computers, despite their abilities to process and convey vast amounts of information, communicate via ·a narrow-bandwidth interface. This interface has remained constant since the 1970s despite an increase in a computer's ability to process user input. Current input devices, such as the mouse, are unable to utilise this increase, thereby constraining the interface between human and computer. The development of alternate input devices revolves around the creation of high-bandwidth interaction techniques. Eye-gaze tracking is one such technique. It utilises the processing abilities of current computer systems to allow new forms of interaction between humans and computers based upon the physical characteristics of human eye movement. Current eye-gaze tracking methods are intrusive. They often require the user to: wear cumbersome equipment such as contact lenses; to use devices such as a chin rest to restrict head movement; or, to operate under special lighting conditions. Such intrusion has restricted widespread mainstream use of these techniques. In contrast, video-based eye-gaze tracking (VBEGT) is non-intrusive. Being able to track eye-gaze from a distance, without making contact with the user, makes this technique attractive to mainstream computer users. Current VBEGT systems use low-resolution cameras to track user eye movements. This study aims to produce an accurate real-time VBEGT system and investigate its suitability for mainstream use. This investigation is based upon the assumption that an increase in both digital camera resolution and speed of image transfer in contemporary technology will realise a significant improvement upon previous VBEGT systems to the point where it becomes practical for mainstream use.