Real-time eye detection and tracking for driver observation under various light conditions

Abstract

Eye tracking is one of the key technologies for future driver assistance systems since human eyes contain much information about the driver's condition such as gaze, attention level, and fatigue level. Thus, nonintrusive methods for eye detection and tracking are important for many applications of vision-based driver-automotive interaction. One problem common to many eye tracking methods proposed so far is their sensitivity to lighting condition change. This tends to significantly limit their scope for automotive applications. In this paper we present a new realtime eye detection and tracking method that works under variable and realistic lighting conditions. By combining imaging by using IR light and appearance-based object recognition techniques, our method can robustly track eyes even when the pupils are not very bright due to significant external illumination interferences. The appearance model is incorporated in both eye detection and tracking via the use of a support vector machine and mean shift tracking. Our experimental results show the feasibility of our approach and the validity for the method is extended for drivers wearing sunglasses.