The Research On Real-time Monitoring Of Driver Fatigue Degree

With the development of economic and social, the following vehicular traffic security problem also more and more receives the common people to pay attention. In a certain sense, driving motor vehicles is more danger than driving the train, vessels and aircraft. Recent studies have shown that:a car accident due to a collision about 30 percent is the driver's fatigue driving. Therefore, many countries are actively carrying out the research work-fatigue driving.This thesis refers to the domestic and foreign literature and studies the condition of the eyes to evaluate the pilot whether the driver is driving in fatigue.Based on the colour-model of face detection technology, this thesis separates colour region from non-colour region, and finally using the "four border connectivity" algorithm of borders, this thesis gets to the external rectangular that included face region. quickly and accurately. In order to improve the speed of human eye detection, this thesis puts forward a new approach to detect the human eye. Combined the Gray projection curve with the Gray histogram estimate the initial threshold value, the test result indicated this approach is fast, simple and accurate, then unifies the Kalman filter and the Mean Shift algorithm to the human eye track. The track effect under the actual illumination is good, regarding covered the eye partly or the close situation can also track accurately. This thesis uses the approach of template matching, and standardizes by similarity measure to identify the situation of eyes. This thesis based on the percentage of the Eyes closed duration time(PERCLOS)and the time of eyes closed duration to judge whether the driver is fatigued.