Detection And Implemention Of Engine Driver Fatigue Based On Facial Image

With the rapid development of urban rail transit in China,driving safety has become a problem that can not be ignored.Fatigue caused by long time driving of urban rail transit drivers is bound to affect driving safety.At present,it is mainly guaranteed by regulations and vigilance buttons,this will not only increase the intensity of the driver's labor,but also limit the driving test.So far,there is almost no study on mature and effective method for fatigue driving detection and warning of urban rail transit engine drivers,therefore,in this paper,a set of on-board and real-time driver fatigue driving detection system based on facial features is developed for the special operating environment of urban rail engine drivers.The system hardware is based on TI's TMS320DM3730 processor hardware development platform,adopt TI's TMS320DM3730 embedded processor with ARM core and DSP core,which have a perfect peripheral IO interface;the storage module is512 Mbytes nandflash core board,512 MBytes DDR400 memory;the expansion board is1 Gbytes 8 bit NAND Flash Samsung K9G8G08 UOM,which supports 32 GBytes SD card;using TVP5150 video capture chip as the video input and output interface,and configuration infrared night vision waterproof CCD camera.The system software uses the Adaboost algorithm to achieve the accurate positioning of the human face,and the AAM matching algorithm is used to detect the eye's characteristic state,finally,based on PERCLOS algorithm proposed by the U.S.Department of Transportation to detecte the fatigue state of engine drivers,the algorithm takes the ratio of eye closing time per unit time as the basis to judge whether the driver is in fatigue state.The core algorithm is divided into two steps: first,transplant and optimize TI's Linux DVSDK software package,using AdaBoost algorithm to detect the face of engine driver;second,for the complex environment of train operation,AAM matching algorithm is used to extract the number of blinks per unit time and the proportion of eyes closed over a period of time,statistic the unit of time eyes closed and the number of eyes open and close,and then judge whether the driver is in a state of fatigue.Finally,by testing and optimizing the video image of the operating environment of urban rail engine drivers,the correctness of the work is verified,which proved that the developed software and hardware system has achieved the effect of engine driver's fatigue driving detection.