Research Of Driver's Fatigue State Detection Based On Video

Driving in case of drowse is a most important factor of traffic accidents. Now many countries are active in studying drowsy monitor.This article design and implements a set of driver fatigue detection system. According to the study of physiological changes about the fatigue, the system can detect eye states by means of machine version and image processing analysis, thus detects the degree of fatigue not affecting the driver. At first, capture color video of drivers in different states and environments, get specified frame image by DirectShow. Then locate driver's face by mixing skin model, morphology and regional growth. Analysis edge and gray information to search possible eye regions, using experiential knowledge to pick up a pair of characteristic regions which are best match. Processes eye's characteristic images with morphological operation, vertical projection, smooth disposal, etc. Distill eye's character vector, which are processed by LVQ network to determine the eye states. Processing continuous images, statistics states based on PERCLOS criterion, to judge whether the driver are in fatigue and give an alarm.Many tests show that the system can detect blinking times and visible pupil area in real time. So it can detect the degree of fatigue efficaciously.