Research And Implementation On Fatigue Driving Real-time Monitoring System Based On DM642

With the development of society, motor vehicles increase quickly. Motor driver fatigue is increasingly reported as a problem concerning the road traffic safety. According to the transport sector statistics, our country is the first that the highest number of deaths in road traffic accidents has been for several years in the world. Drivers' fatigue driving is one of major cause of traffic accidents .Therefore, it's important to research on methods which can effectively detect the state of driver fatigue. Although there are some simple driving fatigue detection methods, but with the vehicle, non-contact, real-time driver fatigue monitoring methodology have not yet been resolved satisfactorily. In this thesis, the video analysis method for driver fatigue detection is researched. Eye cues relating to fatigue are extracted based on the analysis of driver videos to determine driver fatigue and give a warning.The paper make reference to the existing literature in Domestic and abroad, research the fatigue detection method based on the video driver. The paper combines the research project of laboratory: Embedded Remote Video System Based on H.264, designing a system of Research on fatigue driving real-time monitoring system based on DSR The main work done in this thesis is as follows:The framework for detecting driver fatigue is determined. Some algorithms for image pre-processing, such as illumination compensate are presented. Adopt face detection technology based on the color transformation model in non-linear color space, face tracking method based on Kalman filter is implemented to speed up driver face region detection. On the basis of the confirming face region, We taken by coarse-to-fine, and gradually positioning strategy to detect the driver's eyes: Obtain the binary edge of facial features in the facial region by improving horizontal Sobel operator. An improving 8-connected component algorithm is used to repair the part of the fracture edge. Then confirm the location of eyes accurately by integrating facial features' geometrical characteristics, region growing and morphology algorithm. After eye's region localization is obtained, parameters relating to driver fatigue such as the eyes closed time and frequency of driver' blinking are detected.In this paper, we adopt the video analysis method to detect driver fatigue in real time, the method is better to solve the problem at present. To achieve the goal of Real-time Monitoring System, we implement hardware platform based on the TMS320DM642 DSP processor produced by the TI Corporation, profile performance, and analyze opportunities for optimization using CCS. The system which has the small size can be installed in the drivers' cab and it doesn't affect driver's driving. The results of our experiments demonstrate that this system achieve the goal of reducing cost and improve the performance, is better than existing PC based systems.