Research Of Driver Fatigue Detection System Based On FPGA

As the rapid development of the gobal economy and society, there is a constant growing trend of car ownership. But with it comes increasingly frequent road traffic accidents resulting in huge economic losses and life casualties, in which drving fatigue plays an important role. Fortunately, researches show that the great majority of those accidents caused by fatigue can be effectively avoided by means of the timely detection and warning in advance.The driving fatigue detection based on computer vision and PERCLOS(Percentage of Eyelid Closure Time) eye movement has become one of the most promising technologies due to the non-invasiveness and good correlation. In view of the weakness of software processing speed, the hardware processing system based on FPGA(Field Programmable Gate Array) provides a feasible implementation by virtue of its high integration, strong parallel processing ability and abundant hardware resource. Therefore, a real-time driving fatigue detection system is developed and validated by experiments, in which the states of the driver's eye are measured and the value of PERCLOS is computed in real-time for fatigue judement. Meanwhile, all the control and process modules are integrated on FPGA device. The specific studies are as follows:(1) Proceeding from the significance of driver fatigue detection, the exsiting measurement technologies and devices are introduced. By analyzing the superiority of near infrared eye recognition applied in driver fatigue detection and the tendency of signal processing via hardware implementation, a combination of near infrared eye recognition and FPGA implementation is put forward.(2) The overall framework of driving fatigue detection system is developed and near infrared video acquisition setup is focused designed. This setup consists of on-axis LEDs with wavelength 850 nm and narrow-band filter attached to the camera for acquiring the clear and stable near infrared eye imgaes, which enables the system has all-weather and all-time working capability. Meanwhile, the system is implemented on FPFA device produed by Xilinx company with high effiency hardware language VHDL and the synthesis software ISE, realizing the desired functions, such as I2 C configuration, video acquisition, data storage and synchronized display.(3) The key driving fatigue detection algorithm, namely the near infrared eye recognition, is studied. Aiming at the eye target taking only few pixels and lacking texture information, firstly, gamma correction is adopted to increase the contrast; secondly, an eye window is located based on the top-hat transformation; thirdly, a simplified PCNN(SPCNN) model is used to achieve accurate, complete and effective segementation of eyes. Last but not the least, the whole algorithm is transplanted into FPGA.In this paper, the driving fatigue detection system in this paper is constructed on the FPGA platform, which is flexible and portable, low profile, with strong anti-interferance ability and all-weather working capacity. A series of experiments are carried out. The results show that the system can work steadily and correctly and is able to realize the real-time online driving fatigue detection.