SIMULATION RESULTS FOR AN INNOVATIVE POINT-OF-REGARD SENSOR USING NEURAL NETWORKS

A major obstacle in point-of-regard monitoring for human-computer inte raction has been the contaminating effect of head movement. A novel so lution to this problem has been simulated. A multi-layer perceptron co nverts the distorted pattern of four infrared sources, reflected in th e cornea of the user, into an estimate for the point-of-regard. Using an idealised model of the eye the simulation combined vertical, horizo ntal, pitch and yaw head movements with eye excursions typical of indi viduals seated in front of a computer display. Results show that the m ethod is capable of dramatically reducing the effects of small head mo vements. Under these viewing conditions the error in the point-of-rega rd estimate exhibited a standard deviation of under 0.5 mm. I is concl uded that such a scheme could form an attractive solution to the long- standing problem of head movement artefact.