Research Of Driver's Fatigue State Detection Technology Based On Video

Fatigue Driving is regarded to be an important reason of traffic accident. Drivers in the fatigue state, perception of the surrounding environment to determine the capacity and ability to control the operation of the vehicle will be decline. Statistics show our country's traffic accidents also increased rapidly, it became a threat to people's life security as well as social order and stability. In order to protect the tranffic security and to prevent traffic accidents, it is very necessary and meaningful to research effective ways to detect driver fatigue status and fatigue alarm. This dissertation based on analyzing of the domestic and international research status and development trend in the field of fatigue driving monitoring, summed up the current commonly used methods to monitor fatigue driving. Then a driver fatigue monitoring system based on video is presented. The main work done in this thesis is as following:(1)Location of driver's face. In order to improve efficiency, first process by light correction. As the skin color has the good assembling character in the color room, this thesis selects YCbCr color room establish skin color model. Potent factor of face judgment is introduced to further removing non-face regions in the skin color segmentation results, so the searching regions for face detection reduce and the face is located at last;(2)Face tracking based on CAMshift algorithm. In this paper, the face location results as the initial tracking window of the CAMshift algorithm, thus achieving automatic face tracking;(3)On the basis of the confirming face region, the driver's eyes are located and detected. an eye detection method based on Support Vector Machine-Decision Tree(SVMDT) is proposed in this paper, which can achieve the location and the state detection of the eyes at the same time. In the detection, the non-eye region can be eliminated partly based on priori information to reduce the eye region, at first; The eye can be detected through SVMDT, at last. Experimental results show the feasibility of the method;(4)Fatigue recogonition. After accurate positioning of the eye region, calculate the eye frame cures and distance between canthus line connecting to get the open degree of the eye, and by calculating the corresponding PERCLOS and maximum eye closure duration, we can detect the driver's fatigue state.