Research On Driver Fatigue Detection Technology Based On Visual Information Fusion

With the economic development, there is more than1billion vehiclesaround the world,which means that every seven people own a car. Therefore, theoccurrence of traffic accidents is becoming more and more frequent. According tostatistics, the three major cause of traffic accident is drunk driving, speeding drivingand fatigue driving. And technology of drunk driving detection and speeding drivingdetection has been very mature, but fatigue driving detection is always a difficultproblem. In this paper, by using computer vision and image processing technology,first of all to face detection, eye location and secondly on the basis of this and thenget three fatigue characteristic parameters respectively is the PERCLOS(eye closedpercentage),ECT(eye closed time),EBF(eyes blink frequency),the final decisionmethod D-S evidenve theory is used,this a few characteristic parameters, a moreaccurate to determine fatigue state. In this paper,the main research contents are:(1) Face detection which based on skin color information quickly locatehuman face rough area, then use AdaBoost classifier to detecte and locate thehuman face.(2) Eye location which according to the face image, use the method ofgeometry and integral projection coarse positioning the human eye, and thenthrough regional growth accurate positioning the human eye.(3) The fusion of three fatigue parameters by using the method of D-Sevidence theory. the parameters respectively is PERCLOS, EyeClosureTime andEyeBlinkFrequency frequency.(4) Combined with opencv database, written in c++program and verifythe accuracy of the algorithm.A large number of experiments shows, that this paper studies the fatiguedetection based on image information fusion algorithm has good accuracy androbustness, and the speed also meet the requirements, which has a good foundationfor transplanted to embedded devices.