Automated drowsiness detection through facial features analysis

The lack of concentration, caused by fatigue, is the most factor of the increasing number of accidents. In the last few years, the development of an automatic system which based on facial expression analysis, to controls the driver fatigue and prevents him in advance from accidents, has received a growing interest in all intelligent vehicle systems. In this paper, we propose and compare two methods to detect the driver drowsiness state. These methods extracts geometric features using video to characterize eyes blinking as a nonstationary and nonlinear signal. The first method is based on Cumulative Blink Signal analysis technique "CBS" which locates and analyses the eyes blinking from the obtained nonstationary and nonlinear signal to detect the driver drowsiness state. The second method is based on IFD technic "Intinsic Functions Decomposition of the nonstationary and nonlinear signal to analyse the nonstationary and nonlinear signal by using the combination between the two methods: Empirical Mode Decomposition (EMD) and Band Power(BP). For both proposed methods, this analysis is confirmed by the Support Vector Machine (SVM) to classify the state of driver fatigue. The synthesis results obtained by both methods CBS and IFD are discussed and compared to those of the literature.