Research And Application Of Vehicle Fatigue Driving Behavior Detection Algorithm

With the advancement of modernization and the development of express service,the transportation industry has been developing rapidly.The traffic accidents caused by fatigue driving have become a problem that needs to be solved urgently.Fatigue is a normal physiological phenomenon,but this seemingly uncontrollable normal behavior has led to a large number of traffic accidents in motor vehicle driving,and caused numerous social problems.Fatigue driving has become a major threat to the lives of drivers and others,and has gradually become a global problem that needs to be solved.Currently,most of the fatigue driving detection methods are based on the indicators returned by the combination of software and hardware.Facing the individual drivers with differences,the difference between fatigue driving and non-fatigue driving is not obvious,which makes the effect of fatigue driving detection unsatisfactory.So far,no mature and effective fatigue driving test equipment has been developed in China,and the research is still in the experimental stage.Therefore,it is necessary to strengthen the research and development of fatigue driving test products.Paper analyses the characteristics of all kinds of detection system,including the biological electrical signal detection system,such as eye detection system based on perclos is put forward based on the convolution neural network combined with radial basis function(RBF)fatigue driving detection algorithm,using the depth convolutional neural network is used to identify the fatigue driving characteristics,and classification by using radial basis function.Built a fatigue driving simulation system based on TensorFlow environment,driver's driving in all kinds of behavior data collection,data training model,and tested in actual road and traffic environment,make sure the driver is in a different level of fatigue facial movements and the performance of driving behavior,improve the performance of the algorithm.Experiments show that the method proposed in this paper can adapt to complex scenes,realize real-time detection of fatigue driving behavior,and ensure high accuracy.