Computer Vision Research On Fatigue Driving

With economic development, the number of cars is increasing. By cracking down on drunk driving, we reduce traffic accidents by drunk drive. Correspondingly, the detection of human fatigue, we can reduce traffic accidents caused by driver fatigue. Driver for a long time will gradually get into the perception of fatigue. On the surrounding environment, the judgment will have a significant decline in the control of vehicles, their behavior will gradually relax, the risk of traffic accidents will gradually increase. A large number of statistics indicate that driver fatigue caused by traffic accidents, accounting for1/8of the total number of accidents, but it occupies1/6of the personal injury accident. Therefore, studies of driver fatigue detection method to reduce the occurrence of major social significance.This dissertation based on analyzing of the domestic and international research status and development trend in the field of fatigue driving monitoring, summed up the current commonly used methods to monitor fatigue driving. Then a driver fatigue monitoring system based on video is presented. The main work done in this thesis is as following:Face detection and location of the human eye. Summary of the existing face detection and location, we propose a face detection algorithm combining based on knowledge-based methods and based on statistical methods. The use of skin color clustering can be quickly implemented Adaboost algorithm to detect with high accuracy does not depend on the characteristics of the color space to improve the accuracy of face detection algorithm. Face location based on the face of the organ distribution of knowledge and Adaboost algorithm combined with the precise positioning of the human eye.Track face Camshift algorithm. Fatigue testing computer-based visual processing problems will be faced with the problem of lack of real-time. In this paper, an improved target tracking algorithm, and improves overall system efficiency. Optimal adjustment of the process of Camshift tracking algorithm to solve Camshift algorithm, semi-automatic and dealing with problems in the tracking fails.Fatigue measure and the human eye state recognition. PERCLOS algorithm state metric based on human body fatigue the most effective method, but very subjective metric, according to the principle of PERCLOS try to propose a new measure, and enhance the objectivity of measurement. On this basis, this paper proposes the use SVM vector state machine algorithm to identify the human eye state.