Research On Driver Fatigue Detection Technology Based On Eyes’ State

With the social development, traffic safety has become increasingly prominent. As a big culprit in the traffic accident, driver fatigue has brought great disaster to humans. Traditional protection, such as seat belts, airbags, bumpers and so on, do not take preventive measures. At present, the technology about driver fatigue detection has become a hot research topic of vehicle safety driver aids, and can effectively improve the car’s active safety. Therefore, the research is valuable for practical application. After the analysis and summary of the advantages and disadvantages of the existing driver fatigue detection method, this article involves the design of hardware and algorithm of the driver fatigue detection system.1. Aiming at the implementation of driver fatigue detection embedded system, a video system is built based on the DM642processor. The working principle of video acquisition system based on DM642and TVP5150is discussed, and the key technology on the development of video device driver is explored. In the basis of the thorough understanding of FVID class driver model, the DM642driver is designed based on FVID class, and the acquisition of video frame is realized.2. An algorithm of PERCLOS-based driver fatigue recognition is proposed. Using the method that is from rough to fine, the algorithm locates the eyes which can reflect the fatigue state of the driver step by step. The face region is detected by AdaBoost and Haar classifier face detection algorithm in the image of driver’s driving which is obtained by CCD camera. Using the facial priori knowledge, the ROI (Region Of Interest) of the eyes is selected. After a series of related processing, the candidates of the eyes are obtained and verified more precisely, and then the eye is accurately located by analyzing its integral projection curve. Finally, the status of the eyes is identified by calculating the correlation with an opening eye template, and the state of driving fatigue is determined by calculating the value of PERCLOS.3. According to the ideas of the RF5multi-task reference frame, a fatigue detection DSP embedded system is designed, and the various functional modules of fatigue detection algorithm in the RF5framework.