Dsp-based Driver Fatigue Monitoring Methods

The increasing number of traffic accidents due to the drivers' diminished vigilance level has become a serious problem for society. It's pressing to design an embedded system, which can monitor the drivers' vigilance, to avoid the traffic accidents.This paper presents a computer vision system for monitor drivers' vigilance in real time based on DSP. Face detection, eyes position and fatigue estimation are discussed in the paper. The main content of the paper is described as follow.1. The study of face detection and eyes location is worked on the basis of DSP. The background difference algorithm and face-color model are used to detect the face, which locates the face quickly. Aiming at the eyes location, this paper brings forward the method of integral projection combined with region growth to lacate the eyes. In addition, spiral queue is applied here to improve the conventional method of region growth. In this way, it can acquire the state of the eyes quickly and exactly and meet the need of the robust and the real-time of the system.2. The method of fuzzy classifier is used to estimate the drives'state based on the state of the eyes, which integrates the parameters of PERCLOS, ECT and EBF. This method can reduce the uncertainty of the drives'state estimation which may be made by the only parameter.3. A vision system for monitoring a driver's vigilance in real time is designed based on DSP (TMS320DM643) .The system has been tested with different users and results show that the system can monitor the drive's vigilance and estimate the state of the driver.