The Study Of Fatigue Detection System Of Drivers Based On The Eye Status

Nowadays, with the improving of people’s living standard, people of all countrieshave more and more cars, but there are also more traffic accidents at the same time.Many studies show that driver’s fatigue is an important reason for the growing oftraffic accidents. So it is of great significance to carry out driver fatigue detection andpre-warning research.Based on previous work, we consider that the driver’s fatigue characteristics canbe reflected by human eye obviously in my thesis. Focusing on the human eye, wedeveloped a driver fatigue detection system. The whole system is divided into fourparts: locating driver’s eye and iris, tracking iris and eye, identifying the state of thehuman eye and estimating eye gaze direction, and detecting driver fatigue. There aremany parameters in determining the fatigue. We take three parameters, which arePERCLOS, blink frequency, eye gaze direction, to determine the fatiguecomprehensively. Our works are as follows:In the first chapter, the application background of driving fatigue detection andcurrent situation of research at home and abroad were introduced. The significance offatigue detection and the content of my thesis were expounded.In the second chapter, we proposed a effective method for eye and irislocating.At first, we detected the human face by using of Adaboost algorithm. Afterthat, we used integral projection in eye region of interest, and located the eyes. In thefeature extraction part, we extracted contours of the eye, fitted with ellipse, and finallylocated the iris by using Hough transform.In the third chapter, we proposed an improved Mean-shift tracking algorithm forthe human eye. Mean-shift algorithm often tracks failure when target moves fast. Sowe introduced the SSD image matching algorithm, improved Mean-shift algorithm, and took advantage of prior knowledge to track the iris.In the fourth chapter, the state of driver’s eye was recognized. Firstly, theopening and closing state of the human eye was determined by using of eye height.Secondly, blinking frequency analyzed. Finally, we proposed an efficient method todetermine the driver’s gaze direction, which is determining the position of iris in theeye by comparing the center of the eye with coordinates of the iris center.In the fifth chapter, we proposed three driver fatigue detection methods. Basedon PERCLOS measurement method, we proposed the method which is based on blinkfrequency and which is based on gaze direction estimation, respectively.In the sixth chapter, we summarized our work and the objectification research offatigue detection of driver is prospected.