Comparison of different algorithms for robust detection of pupils in real-time eye-tracking experiments

Abstract

This paper contains a comparative investigation on algorithms to find a pupil within an image of a video eye-tracker and to extract the parameters of an ellipse describing the perimeter of the pupil within the 2D image. In future these algorithms shall be used to investigate on the accommodation behaviour of subjects during every-day life activities and to estimate the power consumption of novel, active eye implants like the Artificial Accommodation System. © 2013 by Walter de Gruyter.