Driver Fatigue Detection Based On Dynamic Characteristics Of Human Eye

Since the advent of car, the number of motor vehicles and drivers unceasing increase, lead to the occurrence of traffic accidents number is on the rise. Fatigue is one of the important reasons of causing traffic accident, it has become a major current social problems. When the driver fatigue driving the human body will appear a series of symptoms associated with fatigue, eyes of dynamic state is one of them. So studying a kind of effective monitoring the driver system of the human eye fatigue can effectively reduce the number of traffic accidents, and improve the driving safety.This paper is to study how to rapid and accurate determination of the driver’s eyes open size, thereby further determine whether the driver is in a state of fatigue. This thesis research content mainly includes the following three aspects:1. Analysis of the existing face detection methods, realized face detection in YCbCr color space using skin color model, based on the face detection then use Meanshift algorithm for face tracking. In order to reduce the driving outdoor lighting effects, the illumination compensation method is used to improve the tracking accuracy.2. Using the eyelids "three court five eye" to coarse positioning human eyes in the trace face region. Due to the impact of illumination, in order to effectively locate the human eye, this paper adopted a S (saturation) component in HSV space which not sensitive to light, combine with the illumination angle, extract the weight figure only horizontal component is done in the edge detection, get accurately eye location.3. Using least square ellipse fitting method to extract the eye contour, extract ellipse parameters, and then get the human Eye Open Degree(EOD). To avoid the effects of different sizes of human eyes, the relative size of the human eye is used to judge the driver’s eyes, which is in order to determine whether the driver is in a state of fatigue.This paper realize algorithm function of each part fatigue driving detection, based on the these implements the comparison of all kinds of classic algorithm. Then choice the environment suitable for driving the algorithm, form a complete fatigue detection system. Test results show that the system can detect fatigue, and achieve the desired purpose.