Study Of Driver Fatigue Based On Vision Detection System

With living standards improved,people'S demand for cars increasing Traffic safetyissues become the hot topics The traffic accidents caused by driver fatigue attract moreattention Fatigue affect the driver'S driving behavior that will lead to coy crashes,SO manyresearch institutions aJ'e active in launching the driver fatigue detection and preventionThat design the vision system uesing the DM6437 device as the core,and usedevelopment boards,diodes,filters,image sensor and a series of components to build thehardware platform System software from the function is divided into image acquisition,image processing,voice alarm and video display several modules During debug stage,thedevelopment board is connected with PC through the simulator．Then the program isburned into the DM6437'S FLASH The detection system become an independent testsystemIn the image acquisition module,employ single ring 850nm near-infrared LED lightsource,whose sta~up is controlled through setting the DM6437'S GPIO port Imageprocessing module is the vital module of visual inspection systems mainly for facedetection,eye location and extraction of characteristic parameters ofthe eye According towavelet theory will be the first to scale the image to reduce the amount of data to beprocessed,then the use of coaJ'se-to fine positioning of eyes and ideas to achieve thepositioning,improve positioning accuracy．Firstly,use the subtract background method toget the head outline information,and propose the background update method,secondl~binarization the difference image,then employ vertical and horizontal integral projectionmethod to determine the scope of the face According the face geometric distributionfeatures,narrow the eyes search scope,in which using differential vertical and horizontalprojection to determine the eye'S center coordinates Through mapping the eye'S centercoordinates,devide the eye image from original image In the segmented image,using acustom template detect eyelids,and mark the upper and lower eyelids,then extract themaximum distance value between the eyelids. Analysing the video image sequence,application PERCLOS eye fatigue evMuationprinciple to determine the eye fatigue degree Research the relationship between the fatiguethreshold and the sample rate Laboratory six testers data show-that when PERCLOSvalues near the 15％,the fatigue sign is much obvious,the sample interval less thma 120mscan basically meet the real-time requirements,but longer sample interval will Mt'ect thePERCLOS value,thus Mt'ect the verdict...