Research On Driver Fatigue Detection Based On Eye Information

In today's society,traffic accidents happen every day.And these accidents are mostly caused by the fatigue of drivers.So the research on driver fatigue detection,which takes eye location as its accurate real-time foundation,will play an important part in preventing the occurrence of traffic accidents.the eye location is the basis of accurate real-time fatigue tests.With the help of video image processing knowledge,the status information of human eyes in driving can be acquired.On the basis of the PERCLOS criterion and blink rate,the fatigue driving will be detected through the status information of eyes.Therefore,this thesis designs and realizes a set of driver fatigue detection system,which is based on the combination of PERCLOS criterion and blink rate.The main research works of this thesis are as follows:(1)Using the skin color model under HSV and YCbCr space to detect human face.After analyzing and comparing the detecting results of various commonly used face detection algorithms,it can be found that the skin color model for face detection can overcome the influences of illumination and facial gestures.Moreover,it has obvious advantages both in speed and accuracy in detection.Therefore,the skin color model based upon HSV and YCbCr space is chosen as the face detection scheme.Then it is utilized to analyze the influence of different color space on complexion extraction,and study the process of face detection according to the skin color clustering factors.(2)Optimizing the coupling parameters of Pulse Coupled Neural Network(Pulse Coupled Neural Network,PCNN).Based on the common ways of the human eye detection,this thesis puts forward a PCNN model-based human eye detection method with the optimization of coupling parameters.While the detection rate declines as the traditional methods are used to detect human eyes in closed condition,this method eliminates that decline.And the experimental results show that the human eye detection algorithm put forward in this thesis is accurate and effective in different light conditions,different facial gestures,and closed eyes condition.The human eyes can always be precisely pinpointed in a high speed and precise without being affected by the open or closed degree of the human eyes.(3)Implementing the driver fatigue detection with the coefficient utilization of the blink rate and PERCLOS criterion.Starting with drivers' features in videos,their eye conditions in driving process are taken to judge the degree of drivers' fatigue.After analyzing the limitations of the single usage of PERCLOS criteria for fatigue judgment,the thesis proposes a new driver fatigue criterion that combines the PERCLOS norm with blink rate.Conducting a driving simulation in PC can realize the driver fatigue detection.And the experimental results show that this method can improve the accuracy of driver fatigue detection without wasting more time.