The detection algorithm of locomotive drivers´ fatigue based on vision

To do exact, real-time and non-contact detection of the locomotive driver´s fatigue, an algorithm based on vision is proposed in this paper. In color image of the locomotive driver, the human face is detected by combining the clustering characteristic of skin color in color space and basic human face features. Then, a new method of projection based on mask is designed for locating the eyes. Finally, the PERCLOS is adapted to recognize driving fatigue. The experiments validate the effectiveness and reliability of the algorithm.