Design And Implementation Of Driver Fatigue Warning System

The rapid increase in car ownership and the number of drivers result in increasingly serious traffic problems, and driver fatigue is the main reason for inducing traffic accidents, so accurate fatigue detection and timely warning given to the driver are important for reducing traffic accidents. Because it does not affect the driving operation and has the real-time monitoring ability, the fatigue detection technology based on visual characteristics gradually becomes a trend for driving fatigue detection, but the detection results are affected by changes in illumination and driver’s individual differences, and the accuracy of fatigue characteristics extraction and fatigue detection is not high.Firstly, this thesis studies the existing driver fatigue detection technologies, and on the base of these, designs a kind of driver fatigue warning system based on the principle of visual-based non-contact detection technology, with the driver’s eye features, mouth features, head features and lateral displacement feature when driving. Upon finding fatigue, the system gives warning to the driver by voice and visual interface. The system uses Zedboard FPGA development board as the hardware platform, and by designing image acquisition module, it implements the acquisition of driver’s facial image and road image; by designing image output and voice output module in FPGA, it implements the warning of voice and interface output; by using the SoC technology, it integrates the ARM hard-core processor in Zynq7000, image acquisition module, the image output module and the voice output module, and finally implements the overall hardware design; by transplanting Linux operating system to Zedboard, it achieves multitasking needs, by transplanting Qt graphics library, it facilitates the development of fatigue warning interface, by transplanting OpenCV vision processing library, it implements part of the image algorithm processing; by developing the upper application in the Linux environment, it achieves feature extraction, comprehensive judgment of driver’s mental state and fatigue early warning function, and finally implements the overall software design.The contributions of this thesis include:(1)To the problem that the eye contour is susceptible to eye glass frame in eye feature extraction, this thesis designs a kind of glass frame filtering algorithm according to the connected domain characteristics of eye region and it achieves good filtering results.(2) By utilizing infrared imaging technology in the design of facial image acquisition module, it reduces the influence to the feature extraction in the situation of complex changes in illumination.(3)By using multiple features to comprehensively judge the driver’s mental statement, it avoids high error ratio in fatigue detection results.(4)By using the SoC technology to develop the automotive driver fatigue warning system for product prototyping, it is conducive to the rapid transition and follow-up upgrade.The test results in actual driving environment show that the proposed system realizes the function of fatigue warning. It can find and give warning to nonstandard driving behavior in time. The accuracy of fatigue feature detection is over 80%, and the accuracy of simulated fatigue detection is more than 80% in the existing test conditions. In a word, it has robustness to the detection of fatigue under complex illumination conditions.