A VOG-based Gazing Point Detection Algorithm

Information regarding to eye movement is applied to many fields such as psychology, ophthalmology, physiology, rehabilitation medicine, web design and human-machine interface. So far many technologies have been developed to detect the eye movement and most widely used techniques include EOG (electro-oculograph), Purkinje image tracker, scleral search coil technique, VOG (video-oculograph). This study aims at development of algorithms that can detect and track the gazing point by comparing the relative coordinates of the center of a pupil in the eye image acquired by a CCD camera and image grabber and the center of the four infra-red reference points image reflected from the eye. The size of input image was scaled down to 1/16 to detect the eye and its neighbor areas quickly. An approximative eye position is calculated from this scaled-down image for the detection of precise position of the center of pupil and four reference points reflected from the eye. These four reference points are created by infra-red illumination not to disturb subject's attention to the target. The infra-red lamps were turned on for the time duration of one frame of image and turned off for the next one frame so that the four reference points reflected from the eye could be extracted easily by comparing these two images. The theorem for a circumcenter of a triangle was applied to detect the center of pupil. The gazing point could be calculated by the relative distance and angle between the center of pupil and the center of four reference points in the captured image.