Research On Fatigue Driving Recognition Based On Human Eye State Detection

With the improvement of human living standards,motor vehicle use increased year by year,which accompanied by the occurrence of various types of traffic accidents,safe driving has become one of the concerns of the community.Among them,a large proportion of traffic accidents are caused by fatigue driving.Therefore,it is important to study a set of fatigue driving recognition algorithm which is convenient,efficient and accurate.In all the external features,the human eye state is the most able to respond to fatigue,therefore,this article through the detection of human eye to complete the determination of fatigue driving.In this article,the following four parts of the work are done on fatigue driving identification studies:The first part is image preprocessing,first,the image is automatically white balanced,so that the color of the image is consistent with the real color of the object,easy to computer identification,and then through the Laplacian-based image enhancement,making the outline of the object more obvious and the face is easier to identify.The second part is face detection,on the basis of the previous study of the more mature face detection are based on the Gentle-Adaboost algorithm,the improvement was made.The Gentle-Adaboost algorithm is combined with haar feature recognition to ensure recognition rate,then the concept of mean hash algorithm is introduced,the face information is transformed into hash fingerprint information,the comparison of hash fingerprint information is used instead of face detection,and then use the cache database to store similar images of the hash fingerprint,this not only can get the precise positioning of the face and the average recognition time also reduces the original detection time of 80%.The third part is the human eye recognition,divided into human eye positioning and human eye recognition,this section presents an improved scheme based on finding the minimum threshold to determine the human eye state,using the template matching method to find the human eye position in the fatigue state,and thus shorten the recognition time of the human eye.Finally,the fatigue judgment part,using PERCLOS fatigue judgment criteria.According to this criterion to determine whether the driver is in a state of fatigue,if it is,timely warning of fatigue,reducing traffic accidents.Experiments show that the average time of fatigue testing is 33.29 ms,which is the least compared with the improved algorithm in some literatures.Experiments show that the improved algorithm can effectively detect the fatigue state of the driver,the average recognition time is greatly reduced,and the expected experimental target is achieved.