Design Of Fatigue Driving Detection System Based On Convolution Neural Network

During the long driving process,the driver may feel tired mentally and physically,which will make it difficult for the driver to concentrate,reduce his judgment ability,and fail to respond in time when the driving situation changes.This phenomenon is called fatigue driving,it will bring safety hazards to the traffic and easily lead to road traffic accidents.Therefore,the research on how to prevent fatigue driving is becoming a hot topic now.At present,domestic and foreign researchers are mainly focused on the three core technologies of fatigue driving identification,early warning and control,and their research fields and methods are widely related to safety science,physiology,medicine,behavioral science,automotive engineering,information science,electronic detection and intelligent control,etc.Based on the researches on fatigue driving at home and abroad,this paper studies and designs a fatigue driving detection system.The specific research contents are as follows:Image acquisition and preprocessing.Analyze the components and imaging principles of the camera,and choose the Weixin Vision SY018 HD industrial camera.Study the image blur caused by the shaking or the uneven illumination in the process of vehicle driving.By comparing various preprocessing methods,a better method is raised to perform image preprocessing to improve image clarity.Detection of facial feature points.The feature point detection algorithm based on convolutional neural network is designed,the classic VGG-16 network in convolution neural network is improved and trained by using data set containing face feature points and finetune technology.The trained network can automatically recognize human face and extract human face feature points.Fatigue feature recognition and extraction.The ear principle is used to identify eye fatigue features,and the EAR principle is improved to realize MAR principle to identify mouth fatigue features.The critical threshold of the fatigue state is found through experiments to extract the fatigue features.Fatigue driving detection algorithm design.Design a detection algorithm based on the SVM principle,fuse the extracted eye and mouth features for fatigue driving detection.Carry out the actual car experimental test,and compare the experimental results with different detection algorithms,through the experimental data.The comparative analysis shows that the fatigue detection algorithm designed in this paper can adapt to the more complicated driving environment and accurately carry out fatigue driving detection.Finally,based on the Windows platform,the realization of the fatigue driving detection system is demonstrated.