A non-contact eye-gaze tracking system for human computer interaction

In this paper we introduce a low cost and non-contact eye-gaze tracking system, which is designed for human computer interaction (HCI). The system has little disturbance to the user. Gazing directions are calculated by measuring the relative positions of the Purkinje Image and the pupil based on the corneal reflect method. The system can process gaze directions in real time, distinguish fixations from other eye movements and use fixation messages to communicate with the computer. A throughput of around 18 gaze points per second was achieved. Experiments show that error rates of the eye gaze tracking system used in HCI in this way is greatly decreased. The fixation messages best indicate the user's intention, which makes interaction more natural and convenient.