Mechanism Research On Driver Fatigue Identification Based On HMM

Driver fatigue plays an important role in road traffic accidents.The number of the accident will be reduced greatly if the driver fatigue can be detected and driver can be warned timely.In recent years,researchers have conducted lots of research on the driver fatigue,and a series of achievement has been achieved.However,on the one hand,indicators at the current moment are used as the only factor to detect driver status mostly,ignoring the dynamic feature of the human fatigue mechanism;On the other hand,the assessment of the level on driver fatigue is greatly influenced by subjective factors,lacking of a unified quantitative grading standard.So the driver fatigue is taken as the research object,and the identification mechanism and detection methods on driver fatigue are researched deeply after generating characteristic is analyzed.Finally,an effective detection method on driver fatigue based on HMM is proposed in this paper.The specific contents are as follows:(1)The causes of driver fatigue and identification mechanism are analyzed.The driver fatigue has dynamic feature by comparing the level of fatigue and continuous driving time.And it is found that drivers’ internal physiological indicators and appearance are distinguishing under different fatigue status.So,the model based on HMM with dynamic feature and double random is built as the basic algorithm framework of the paper.(2)The experimental scheme and a database are designed according to identification strategy.Firstly,the driver fatigue experiment is designed based on HMM theoretical framework,in which the hidden and observed indicators can be obtained.On the one hand,the EEG signal known as ‘gold standard’ in the fatigue detection is selected as the hidden indicators to meet the accuracy requirement,on the other hand,the eye feature is chosen as the observed indicators to achieve mobile applications in the vehicles.Finally,a database containing eye feature and EEG is built.(3)The driver fatigue identification model is built based on HMM.The preferred clustering results of indicators on eye feature and EEG are determined by FCM and Mixed-F,which determine the preferred levels on hidden and observed indicators of HMM at the same time.Then the driver fatigue model based on HMM is verified with driver fatigue database,and the accuracy rate reaches 82%.Focusing on the key issues in the study of driver fatigue,the research of identification mechanism and detection methods are carried out.The points are as follows: Firstly,the eye feature is obtained from DAN which is more robust,then the preferred levels of different indicators are found.And both of the points can improve the detection accuracy of driver fatigue based on HMM.