Abstract
The increasing number of accidents is attributed to several factors, among which is the lack of concentration caused by fatigue. In This paper, we describe the approach developed to detect the driver's drowsiness state from a video-based system to alert him and also reduce the number of accidents. Our approach uses a noninvasive method which excludes any human related elements. The latter calculates geometric descriptors. We analyze the signal extracted from the previous step by combining the two methods EMD (Empirical Mode Decomposition) and BP (Band Power). This analysis is confirmed by the SVM (Support Vector Machine) to classify the state of alertness of the driver.