Eyes location by regional features and knowledge matching for driver fatigue detection

Abstract

This paper presents a novel method for driver's eyes location by combining the regional features and knowledge matching technology. First, the eyes candidates are obtained by the driver's skin color. Then the geometry, gray and location features are used to get the rough location. Finally, the prior knowledge of face structure is employed to decide the precise location. The experimental results show that the new method is efficient and robust to images in different rotation angles. It can work effectively with no limits to the open eyes, two eyebrows or two eyes. The mismatching of the eyebrow and eye on the different side can also be decrease.