Fatigue Driving Monitoring Technology Based On Fusion Feature Of Eog And Imague Information Based On GELM Method

Fatigue driving behavior is a dangerous factor affecting modern road traffic safety driving.The development of fatigue driving monitoring methods will help to improve road traffic safety and reduce the accidents caused by human factors.At present,countries around the world are committed to studying various monitoring methods based on driver fatigue and have made some achievements.The traditional single feature source fatigue monitoring method generally has the characteristics of less feature quantity and strong manual intervention,which affects the driver’s normal driving operation while realizing the fatigue driving monitoring function.Therefore,this paper integrates the characteristics of eye image and electrooculogram,designs a fusion feature driving monitoring system.The feature source of the system is richer than that of a single eye electrical feature and can supplement the key information what is difficult to monitor such eye electrical features as long-time eye closure to improve the misjudgment rate.Firstly,through the analysis and research of classification methods in the field of fatigue driving,the implementation principle of fatigue driving monitoring system is analyzed,and the classification method GELM is selected.Under the condition of fully understanding the principle of multi-source feature fatigue monitoring,the fatigue driving monitoring system is designed and the process of EOG-image feature fatigue driving monitoring system is described.The key problems,related design schemes of fatigue driving monitoring system and fatigue driving judgment criteria are described in detail.Secondly,according to the design scheme of fatigue driving monitoring system,relevant fatigue driving experiments are designed and carried out,and the data required for the experiment are collected.Then,after the signal preprocessing process of EOG and image,various signal processing methods is discussed.In addition,segmentation method based on gradient is used to extract the eye image features.After normalization and windowing,according to the different characteristics of different eye movement information in eye movement signal,the EOG is divided into three kinds of eye movement signals by k-nearest neighbor method,and the relevant features are extracted.For the separated eye image,the co-occurrence matrix gradient information is collected.Finally,this paper discusses the implementation principle of multi feature fusion mechanism,and finally selects concat fusion method.The performances of classification under different feature fusion methods are discussed which prove that the fatigue driving monitoring system used in this paper has good fatigue driving monitoring ability.