A Fatigue Driving Detection System for Train Driver Based on Head Pose Features

Yongda LI,Chao ZHANG,Lingjun MENG
DOI: https://doi.org/10.3963/j.issn.1674-4861.2014.05.020
2014-01-01
Abstract:Fatigue driving of the train driver is a major threat to safety.To overcome the shortcomings of the traditional eye detection methods,this paper presents a new method of fatigue driving detection for train driver based on head pose features.Firstly,the method applies the AdaBoost algorithm to detecting the face region.Secondly,Camshift algorithm is applied to detecting the face and computing the rotating angle of the face.Then,it computes rotating angular velocity and angular acceleration of the head.Finally,it combines the head's rotating angle and rotating angular acceleration to judge the driver's fatigues.A head spin physical model is set up,which can draw the curve of rotating angular velocity and angular acceleration along with the change of rotating angle when the head is freely rotating.A detection system is developed based on the above method.The success rate of fatigue detection is 87.5%;however,it can only detect slow tilting and sudden tilting of the head to the two sides,and is not able to identify yawning,knocking of the head and eye movements,etc.Therefore,it should be combined with other methods for an effective fatigue detection.
What problem does this paper attempt to address?