Researh On Fatigue Driving Detection Method And Its Implemention In Embeded System

As fatigue driving is often the major cause of traffic accidents, so to find a vehicle-mounted, real-time and objective fatigue detection device is becoming the most popular research directions in the current field of automotive electronics among domestic and international researchers. Based on the analysised and summary to the domestic and abroad research about the formation mechanism and detection methods of fatigue driving, this paper analysised and researched the changes states of fatigue driver’s eye movement signals and EEG, and based on these two physiological variation, this paper realized fatigue driving detection and warning system.This paper designed and achieved a fatigue driving detection and analysis system based on DSP developed by the Beijing Ruitai innovative company’ ICETEK-DM6437-B as a hardware development platform, using CCS3.3as the software platform to simulation and optimization algorithms, and using the real-time operating system DSP/BIOS to real-time scheduling and debugging. The main work in this paper is as follows:(1) The module of eye state detecting and processing:as the chroma information has a very good clustering character on the YCbCr color space, so, firstly this paper collected color samples and extracted initialization threshold range of chroma in YCbCr color space; And then this paper did image preprocessing to the real-time image of the head of the driver using median filtering and image sharpening algorithms and located people’s face, using adaptive threshold range of chrominance information. After that, this paper used etching method to help to locate the face precisely. On the base of accuate face position this paper used the geometric features algorithm to locate people’s eyes. Bacause the grey level of eyesbrows are low, it is easy to find the position of the eyesbrows and then to locate the people’s eyes precisely. Because the eyeball is black, this paper determined whether the eye is open or closed by the ratio of black pixels in the eye region. This paper uses PERCLOS algorithm to determine whether the driver is tired after determining the state of the eye.(2)The EEG detecting and processing module:at first, using the EEG amplifier UE-16A produced by SYMTOP instrument Co. Ltd.to to collect EEG; and since the frequency of EEG is mainly below30Hz and its background noise is strong, so using the FIR low-pass filter with Haiming window to approach its filtering. And then analysising the EEG in frequency domain after FFT transformation; through analysising the changing results of the the average power among θ-waves, α-waves and β-waves, each band power percentage in total power, the power ratio of each band and the power-added ratio when the driver changes from sober state to fatigued state, this paper finally finds that it is the most effective method to determine whether the driver is fatigue or not based on (θ+α)/β ratio of the power-added. Finally this paper defines threshold to judge whether the driver is fatigued.Through hardware tests, it shows that the system can effectively achieve the detection and warning function to the fatigue state of the driver; the detection rate is higher, and has good real-time detecting and robust character. So it has a wide application prospect.