Performance of a computer system for recording and analysing eye gaze position using an infrared light device

Our aim is to develop a computer system in order to study the visual scanning and eye fixation during exploration of image. The system uses an infrared device to detect the horizontal and vertical eye movements. In this paper, our interest is concentrated mainly on the calibration procedure. More precisely, on the correction of the nonlinearity in the oculomotor for individual subject's eyes. An adaptive calibration algorithm is developed. The algorithm uses a polynomial model and the optimal correction is obtained by using a mean square error criterion. The paper presents also a method to correct head movements. Experiments show that the proposed approach is fast and accurate for the study of visual perception and recognition processes.