Robust algorithm for pupil-glint vector detection in a video-oculography eyetracking system

This work presents a robust real time algorithm for an eye tracking system employing the well-known bright-pupil technique that performs an effective detection of the pupil and glint positions in the image. The accuracy in the processing is essential if a good determination of the eye gaze is desired. This algorithm is competent for unconstrained images and presents an unmatchable behaviour for users using glasses. The algorithm employs mostly commands from Matrox Imaging Library (MIL) that present a wide sort of functions for image processing and pattern recognition.