Improved Detection of Pupil and Corneal Reflection Using Near-Infrared Light Sources and High-Speed Video Camera

概要

論文の詳細を見る
The position tracking of the pupils and corneal reflections (features) from the video camera images that capture the entire human face is important to precisely and remotely detect eye-gaze. However, doing this was difficult under intense illumination even when using the conventional two light sources and image difference method. In our former study, to resolve the difficulties, bright and dark pupil images and their corresponding non-lighting images were consecutively obtained while turning two light sources on and off.A new image processing algorithm was proposed. In the present study, a high-speed (2000 fps) camera was introduced to decrease the intervals of time need to capture the three images. In addition, the near-infrared light sources were improved to increase the brightness of the small pupil in the bright pupil image. The newly developed system enabled the detection of the features when the subject was wearing glasses and moving their head quickly (20 cm/s) and there was intense illumination (40,000 lx).