点击次数：

论文类型：会议论文

发表时间：2013-10-12

收录刊物：EI、CPCI-S、Scopus

卷号：457-458

页面范围：944-952

关键字：eye detection; driver fatigue; PERCLOS

摘要：One of the important causes of traffic accidents is driver fatigue. In this paper, a new real-time non-intrusive method to detect driver fatigue is proposed. Firstly, face region is detected by AdaBoost algorithm because of its robustness. Then a region of interest of the eye is defined based on face geometry. In this region, eye pupil is precisely located by radial symmetry transform. With principal component analysis(PCA), three eigen spaces are trained to recognize eye states. Open, closed eye samples and other non-eye samples in the face region are used to get these eigen spaces. At last, PERCLOS and consecutive eye closure time are adopted to detect driver fatigue. Experiments with thirty two participants in realistic driving condition show the reliability and the robustness of our system.