Driver Fatigue Monitoring System And Achievement Based On Face Detection

According to the<Report of Chinese Automobile Developing2012-2013> from Chinese Academy of Social Sciences,the number of private cars has exceeded a hunred million. With the improvement of people’s living standards,safety of traffic has become more and more important.According to the statistical result,about25percent to30percent of the traffic accidents are caused by fatigue driving.So how to check if the driver is in the state of fatigue effectively,and lower the threat from fatigue driving has become a hot issue of automobile safety.The thesis has achieved the target of checking the fatigue state. At first,it analyse the current checking techniques and their theory.By comparing the advantages and disadvantages,we found the system based on Face Detection has theproperties like reakime,no contacted and high precision.then Ianalysed and compared lots of materials of fatigue driving checking function,and selected the PERCLOS method as the most precise method related to real state of fatigue driving.The method checks whether the time of eyes closed has reached some percentage of all time or not.Its key is to locate eyes and cacuulate the height of eyes,which takes Face Detection as the premise.To improve the precision and real-time properties,the thesis used ASM(Active Shaped Model) to lacation the face,then separate the zone of eyes by5characteristic points.After that,process the zone with auto-adaptiive threshold method,and get the Characteristics arguments of black connected region to locate the eyes precisely and obtain the arguments of eye.After I improved the fatigue driving checking method,this thesis developed the real system based on PC and Microsoft Visual Studio2008,It mainly includes five part like obtaining image in real-time,face detecting,eyes locating,judging of eyes state and fatigue state checking.Then did some testing analysing of the system.According to the result of experiments,the system can check the fatigue state of driver effectively.The accuracy has reached90percentage,which is very high.