Pupil detection algorithms for eye tracking applications

In this paper two pupil detection algorithms (based on Starburst and circular Hough transform) for the eye tracking application are presented. The operation of the eye tracker device depends mainly on the infrared video camera performances, PC processing power of the video signals and the precision of the pupil detection algorithm. Both algorithms used for the eye tracker implementation have been comparatively analyzed in different conditions for a set of subjects in the field of assistive technology. The performances of the proposed system, given by algorithm running time, cursor stability on the user screen and algorithm accuracy, have been presented.