An Iris Outline Tracker: For Various Eye Shapes in a Single Video Camera without Infrared Illumination

This paper describes a sophisticated method to track irises in a monocular video sequence without infrared illumination. In order to reduce the influence of eyelids and to cope with the variance of the appearance of eyes of different people, we propose a variable circular sector separability filter (VCSSF), and use it to estimate the likelihood of hypotheses in the particle filter. We confirmed that our iris outline tracker has the ability of tracking various eyes for different people whose head motion including translation, 3D rotation and zoom, and even when they are wearing glasses. Through the comparative experiments of the VCSSF, the conventional circular separability filter (CCSF) and the ellipse separability filter (ESF), we confirmed that our iris tracking algorithm is more robust than others.