Real-time image-based driver fatigue detection and monitoring system for monitoring driver vigilance

Abstract

In order to realize automatic on-line monitoring of driver fatigue state, an automatic driver fatigue state early warning system based on vision on-line real time detection is established by analyzing driver's eye and mouth states. Firstly, this system use VJ detector algorithm to detect the face, and then in the face region of interest (FROI), MB-LBP feature is used to find the eye region and locate eyes' area rapidly and accurately in the upper FROI. Then Kalman filter algorithm is adopted to track the eyes and mouth. After this feature enhancement and ellipse fitting for human eye image is adopted after the edge points of human eyes, and a threshold is set to match mouth feature such as open, close and yawning, which is used judging the mouth state. Finally, the threshold is set to determine the human eye state by calculating the ratio between short axis and the long axis of the ellipse. Experimental results indicate that the method used can detect the position and states of human eye and mouth accurately and rapidly in the case of different angle and shielding rotation, and the detection rate is higher than 95%. The established driver fatigue warning system can meet the real-time requirement of the driver fatigue state detection.