Research On Driver Fatigue Detection Method Based On Facial Behavior Analysis

With the growth of traffic accidents,this has brought serious losses to the state and personal property.Research shows that the fatigue driving is one of the main causes of traffic accidents.In recent years,many countries already pay attention to the driving safety problems.Driver fatigue detection research has the vital significance.Fatigue detection methods based on computer vision is non-intrusive and real-time.The state recognition of eye and mouth is the key issue for fatigue detection,but it would be affected by changing of illumination and wearing glasses.To solve above problems,this paper designs an infrared video acquisition system for driver and proposes a fatigue detection method based on facial behavior analysis,the method includes four parts:face detection and tracking,extraction the area of eye,state of eye recognition,fatigue detection.First of all,it detects the driver's face based on AdaBoost,which combined with the KCF(Kernelized Correlation Filter)tracking algorithm for fast face detection.It detects facial landmarks from the video image followed by extracting the eye and mouth regions,and then detecting the state of eye and mouth based on CNN(Convolution Neural Network),eventually calculating PERCLOS(Percentage of Eyelid Closure Over the Pupil Over Time),blinking frequency and yawing frequency to detect the fatigue.The experimental results show that the method detect state of eyes and mouths more accurately when wearing sunglasses,then fatigue parameters can be computed more accurately.By combining multiple fatigue parameters,it is more accurate compared with the methods that used single parameter.