3D Model-Based Gaze Tracking Via Iris Features With a Single Camera and a Single Light Source

Abstract

Traditional 3D gaze estimation methods are usually based on the models of pupil refraction and corneal reflection. These methods typically rely on multiple light sources. The 3D gaze can be estimated using single-camera-single-light-source systems only when certain user-dependent eye parameters are available a priori, which is rarely the case. This article proposes a 3D gaze estimation method which works based on iris features using a single camera and a single light source. User-dependent eye parameters involving the iris radius and the cornea radius are user-calibrated. The 3D line-of-sight is estimated from the optical axis and the positional relationship between the optical axis and the visual axis, and then optimized using a binocular stereo vision model. The feasibility and robustness of the proposed method are assessed by simulations and practical experiments. The system configuration required by the method is simpler than that required by the state-of-the-art methods, which shows significant potential value, especially with regard to mobile device applications.