Research On Driver 's Fatigue State Detection Based On Human Eye Detection

Currently, road traffic accidents have become primary factor causing the non-normal human deaths. Every country pays high price for lives and leaves huge economic losses annually. In China, the incidence of accidents has been the highest all over the world in recent years. Fatigue driving is one of the main factors causing traffic accidents. Therefore, how to detect fatigue driving effectively and warn to the driver timely are significant for reducing the accidents because of fatigue driving.In this paper, the major methods of domestic fatigue driving detection were analyzed and compared with. The driver fatigue detection was implemented based on Adaboost algorithm and PERCLOS algorithm. In the process of eye detection,according to the degradation of the traditional AdaBoost algorithm and no distinction between positive and negative samples in the difficult samples,the algorithm introduced samples state identification mark and used a new sample weight updating method so that the detection rate of AdaBoost algorithm was improved. In the process of eye state detection,through the traditional PERCLOS algorithm a fatigue detection method based on dynamic PERCLOS auxiliary and blink frequency was adopted.The blink frequency was calculated in unit time using dynamic PERCLOS value with time’s changing, two parameters was used to determine the driver’s fatigue state and the accuracy of detection algorithm method was improved.Driver fatigue detection process of this paper is face detection,eye detection,eye state detection and fatigue state detection.In the process of face detection, AdaBoost algorithm with Haar-Like features and integral image quick calculation were used. On the detected face region, according to the structure of "three chambers and five hole", the approximate region of eye was obtained easily. The human eye was detected using the improved AdaBoost algorithm method. The eye state detection is through integral projection method to extract information on the eye.Then,the eye of the aspect ratio is obtained, and it determines the state of the eye is opened or closed. The fatigue state was detected using the improved PERCLOS algorithm method.Finally, the above methods were verified using the fatigue state detection system. The experimental results showed that the method can detect the driver’s fatigue quicklyThe average accuracy of fatigue detection has been significantly improved. It has a better detection effect and meets the needs of fatigue detection.