Low Complexity Pupil Tracking for Sunglasses-wearing Faces 2 for Glasses-free 3D HUDs
- Applied Sciences
This study proposes a pupil tracking method applicable to drivers both with and without 11 sunglasses on, which has greater compatibility with augmented reality (AR) three-dimensional (3D) 12 head-up displays (HUDs). Performing real-time pupil localization and tracking is complicated by 13 drivers wearing facial accessories, such as masks, caps, or sunglasses. The proposed method fulfills 14 two key requirements: low complexity and algorithm performance. Our system assesses both bare 15 and sunglasses-wearing faces by first classifying images according to these modes, then assigning 16 the appropriate eye tracker. For bare faces, with unobstructed eyes, we applied our previous regres-17 sion algorithm-based method that uses scale-invariant feature transform features. For eyes occluded 18 by sunglasses, we propose an eye position estimation method: our eye tracker uses non-occluded 19 face area tracking and a supervised regression-based pupil position estimation method to locate 20 pupil centers. Experiments show that the proposed method achieves high accuracy and speed, with 21 a precision error of < 10 mm in < 5 ms for bare and sunglasses-wearing faces both for a 2.5 GHz CPU 22 and a commercial 2.0 GHz CPU vehicle-embedded system. Coupled with its performance, the low 23 CPU consumption (10％) demonstrated by the proposed algorithm highlights its promise for imple-24 mentation in AR 3D HUD systems.
- Appl. Sci. 2021, 11, 4366