Fatigue Driving State Analysis Based On Blink Interval Time Series

With the development of science and technology and the improvement of people's living standards,there are more and more vehicles on the road,and the traffic accident rate raise the same time.Fatigue driving is one of the main causes of traffic accidents,regardless there are clear regulations prohibiting fatigue driving.Contribute to the hiddenness and subjectivity of fatigue driving,it is difficult for the traffic control department and even for drivers themselves to find out fatigue.Therefore,the research of fatigue dri.ving detection technology is of great significance to not only for academics but also for society.In this paper,a combination of embedded systems,image processing,machine learning algorithm and deep learning algorithm is designed as a fatigue driving detection system to realize fatigue driving detection based on the blink interval time sequence.The fatigue driving detection system contains both a hardware part and a software part.The hardware part is composed of a control module,a detection module,an alarm module and a simulated driving module;the software part is mainly composed of a face detection algorithm,a human eye positioning algorithm,a machine learning classification algorithm and an LSTM network.The detection algorithm start with a human face database,as well as training the face classifier,recognizing the face.The Hough transform was used to locate the human eye to collect the eye-eye interval time.Finally,dynamic time warping algorithm and K-nearest neighbor algorithm were used to find the correlation between the blink interval sequence and the fatigue driving state.After the strong correlation is obtained by calculation,the LSTM network model is trained on the blink interval data,and the model is designed for day and night.The model realizes real-time detection of driver fatigue state in different time periods.The experimental results showed that the system meets the requirements of lightable and portable,easy to install,non-contact,it is verified that the blink interval time is strongly correlated with fatigue driving state,and the purpose of real-time fatigue driving detection is realized by using this feature.