2019 年 55 巻 8 号 p. 491-498
In eye tracking methods, near-infrared image or color image has been used separately. In this research, we propose a hybrid eye tracking method using an RGB-IR camera, which can capture a color image and a near-infrared image simultaneously, to compensate for the weakness of each technique. Pupil detection can be worked better than iris detection because the eyelid often occludes iris. On the other hand, iris area can be used for model fitting because iris size is constant. Therefore, the hybrid eye tracking method works as a glint-free model-based approach in our proposed method. The model fitting can be automated as initialization, and the relationship between the 3D eyeball model and eye camera is solved. Additionally, 3D eyeball pose can be estimated without using first Purkinje image, and the pupil size can be measured quantitatively. We evaluated the accuracy of the estimated point-of-gaze and the pupil size, and we confirmed the feasibility of the proposed method through experiments.