Low cost mobile eye tracker for natural illumination conditions.

Abstract

Eye Gaze Trackers (EGT) are devices that estimate the user's gaze position at each instant. Although they are commonly used in psychophysical and neurophysiological experiments, its potential for interaction with computers has been explored. Among the technologies used for eye tracking we can cite electroencephalography (EEG), contact lenses with coils and techniques using video cameras. The use of cameras allows remote eye tracking, i.e., without any device in direct contact with the user. This facilitates setup and use of the EGT, and also provides greater comfort during prolonged use.Our group already has extensive experience in developing remote EGT, using a single camera and one or more external illumination sources. The external source facilitates the processing of the eye images, and also creates reflections on the cornea that can serve as reference points to estimate the gaze direction. When the surface under observation is limited to a computer monitor, the direction can be mapped to the display coordinates using functions computed from calibration.The introduction of micro-cameras facilitated the development of mobile head mounted EGT using a second camera for capturing scene images. The scene images are used to project the gaze information. This project proposes the development of a low cost mobile head mounted EGT, in order to make them more robust to variations in illumination conditions. We we also research new ways to increase the accuracy of the system by correcting the parallax caused by the distance between the scene camera and the eye.The student, Andrew T. N. Kurauchi, has an excellent academic record. He has recently graduated from USP, where he obtained a B.Sc. in Computer Science. His final overall average grade was 8.66, and he has been working in this project since April 2012.