Measure and analysis of a gaze position using infrared light technique

We are interested in identifying the object or area that attracts the subject's attention in images. A computer system is developed to measure and analyze the eye position. Eye movements are recorded using an infrared light device. Our interest is concentrated mainly on the calibration of the system, more precisely, on the correction of measured data according to user's characteristics. Higher polynomial transformation is used to model the mapping between eye coordinates and image coordinates. The mean quadratic error criteria is used. This paper presents how our model works to correct the non-linearities in the oculomotor system for individual subject's eyes. It also proposes a method to correct head movements.