Research Of Real-time Driver Fatigue Detection System Based On ADSP-BF533

Driver fatigue detection system is an important component of the active safety warning system. The research of driving fatigue detection on how to reduce the incidence of traffic accidents is of great practical significance. With the rapid development of DSP technology, face recognition technology has been widely used. Based on the theory of face recognition, this thesis designs an ADSP-BF533-based system, which is used for monitoring driver's eye condition and determining the status of driver fatigue, according to PERCLOS principles of physiology.The research status of the fatigue driving is described in the thesis, and the method of eyelid closure, PERCLOS, is emphasized. After comparing various of face detection algorithms, the method of the clustering of skin color is used in the system, the function of human face detection and location is realized. Kalman filtering is used for tracing face-region. The gray projecting integral method is used for eyes location. On hardware design, the ADSP-BF533 is applied to build the hardware platform. The platform is composed of video capture module, video storage module, keyboard and audio guard module. As is shown in the experiment, the system is capable of detecting the degree of fatigue accurately with a better real-time performance.That the ADSP-BF533 is selected as a core processor exhibits the design requirements, such as, miniaturization, efficiency, integrated. In the thesis, the approach of human face tracking and DSP Program Optimization is used for improving System performance, and propose the future direction of further research.