Researchers led by Aldo Faisal, PhD, from Imperial College London (www3.imperial.ac.uk) have developed an inexpensive eye movement tracking system. The GT3D device consists of two PlayStation Eye cameras attached to a pair of glasses, which capture images of the irises that can then be used to analyze the position of the user's pupils.
To increase the contrast between the pupil and iris, camera optics were modified for infrared (IR) imaging by replacing the IR filters with an exposed and developed film negative that acts as a low-cost IR-pass filter. Two IR LEDs aligned off-axis to the camera were then used to illuminate the eyes of the subject.
Once images are captured, they are thresholded to create binary images from the grayscale data then filtered to remove noise. The center of the pupil within the iris is then calculated by fitting an ellipse to the iris' contour and extracting the x and y coordinates of the center of the ellipse. Pupil positions extracted from the images of the eye are also used to determine the gaze of a viewer.
The GT3D system allows users to click on an item on the screen using their eyes, instead of a mouse. This problem has previously been resolved by requiring users to stare at an icon for a prolonged period or blinking. However, since blinking is part of natural behavior, it happens unintentionally. Instead, the researchers calibrated the system so that a simple wink would represent a mouse click, which only occurs voluntarily.
To demonstrate the effectiveness of the eye tracker, the researchers asked subjects to play the video game Pong. In the game, the subjects successfully moved a bat to hit a ball bouncing around a screen -- a feat that, the developers claim, is difficult to accomplish with other systems.
