Research Of Technique For Driver's Fatigue Supervising Based On Eye Detecting

With the development of highway and the increasing speed,the traffic accidents worldwide each year are raising day by day.According to statistics,25% -30%of the traffic accidents are because of fatigue driving.Thus,a fast,accurate,and real-time driving fatigue detecting system has become the core problem of the field of Intelligent Transportation Systems.In order to detect the fatigue driving accurately the paper analyzed the mechanism of driving fatigue and current detecting methods at home and abroad firstly.By analysis and comparison,the paper selected a good real-time,accurate detecting and strong anti-jamming detecting method-PERCLOS detecting method.The paper made further transformation for PERCLOS detecting method in accordance with the specific situation of research as follows:changed the ratio of time into consecutive frames.The key of PERCLOS detecting method is positioning the human eyes accurately,then judging the degree of eye opening whose premise is face positioning.The paper used weighted mean to change a color image into a gray image firstly, then fixed face window through the edge detecting method-Robert edge detection.The method can position the face roughly,improve the detecting speed and ameliorate detecting effect greatly.The positioning of the human eye is the key of driving fatigue detecting.The paper searched a number of suspicious connectivity areas of the human eye using the regional search for a black-chip connectivity method in the face window area of binary image. Then in these areas positioned the human eyes accurately in accordance with the structure of face features,such as the location of the human eyes in the face window,the shape of the human eyes and symmetry.According to the experimental verification,the accuracy reached 92%.And the method is also accurate to one who wears glasses.The paper developed a detection system of fatigue driving based on eye detecting. The system realized with PC and general camera in the laboratory environment, including facial image capture,face detection,eye detection and the detection of eyes' status.Through experimental verification the system can detect real-time status of the drivers' eyes as well as fatigue.It can also issue a warning fatigue at the same time.The system has a more accurate calculation and meets the needs of video information processing.At last,the paper summarized the major work and analyzed the future pending research.