Terence Tufuor
Senior Design Technical Report



The birth of computers, especially personal computers introduced the need for human-computer interfaces. We needed a way to interact with these devices in a simple and intuitive manner. Input devices like the keyboard, mouse and pointer have made it very easy to send instructions to these devices and make them do what we need them to do; whether is work related, entertainment etc. But what if it could be simpler? What if it could be more intuitive? What if we could instruct these devices just by looking at them?
Eye positioning and tracking techniques have numerous applications beyond our interaction with our personal computers. They are used in numerous disciplines including medical research, psychology, cognitive science and marketing research. We can track eye movement to detect fatigue in workers and drivers, allows merchants to observe what users look at when they are on their (the merchant's) website and even what athletes focus on when playing a sport.
It is important to note that eye tracking is not a product of the computer age. It is something that has fascinated mankind for centuries and something that people have been investigating for centuries. Historians can trace the beginning of eye tracking research all the way back to the 1700's. But as with most things, our progress in this area has accelerated as the available technology has advanced and with the advancement of artificial intelligence and supercomputers, eye tracking has come a long way in the past few decades.


Eye tracking is the process of measuring either the motion of a person's eye relative to their head or the point of vector of their gaze [1]. The eye is usually tracked based on an image that is captured and processed. There are three main activities involved in this process; detecting/locating the eye on the image, tracking the eye's position from image to image, and analyzing the region of the eye to determine the point of gaze. Eye tracking is used either in diagnostic purposes (e.g. What people focus on in commercials, or what drivers pay attention to on the road) or interactive purposes (e.g. Controlling the mouse on your screen).
The tracking system is made up of a camera, IR emitters and a computer for processing the images (we'll refer to the computer as "the system"). Even though passive light can be used in conjunction with a CCD camera, it is better to use an active light source and eliminate all ambient light. [5]. A lot of research has been done to come up with the components to make up the ideal system. Research comparing webcams with CCD cameras shows that webcams are too bulky for these purposes and provide more information than we need [4]. Ambient light is undesirable because it is not always consistent and the amount by which is varies due to time and location cannot always be controlled. It is better to have consistent and controllable surrounding conditions because it improves the chances of success and reduces the amount of research and algorithm development involved. When using active light, IR is preferred to visible light because IR is not visible to humans and so it's less distracting. However, it is not advisable to expose the naked eye to IR light for long periods of time as this can cause damage to the eye. It is recommended that a person uses an IR illuminations scheme with the wavelength around 780 nm. This is below the threshold for human eye safety [1][5].
The IR emitters are arranged in 2 concentric circles with the camera in the center. One ring is small and close to the camera while the other one is farther away from the camera. This setup is used for both eye detection and eye tracking.

Fig 1. The setup for the tracker.


When IR light falls on a person's cornea, part of it is reflected towards the light source and a glint appears on the person's eye. If the person is looking at the light source, the distance between the glint and the center of the person's pupil is small and this distance increases when