Eye state tracking for driving is the key problem in fatigue detection, and it has been proved to be effective and popular method in fatigue driving. In this paper, we present a new algorithm framework for fatigue dri...
详细信息
ISBN:
(纸本)9783642549243;9783642549236
Eye state tracking for driving is the key problem in fatigue detection, and it has been proved to be effective and popular method in fatigue driving. In this paper, we present a new algorithm framework for fatigue driving. We first recognize human eyes with AdaBoost algorithm, then an improved otsualgorithm is modified to automatically adapt to varied environments. Furthermore, we propose an effective algorithm based on dynamic sliding window in order to compute proper threshold between open and close eye window. Finally, we estimate the different level of fatigue driving with improved percentage of eyelid closure time (PERCLOS) algorithm. Human eye images are captured with camera in real time, and our method is simulated on personal computer. The comparative experiments demonstrate that the proposed algorithm framework can effectively discriminate the level of fatigue state in driving.
暂无评论