Research On The Detective System Of Locomotive Driver’s Fatigue State Based On Video Signal

As one of the main mode of world-wide transport, the economic developmentrole of rail transport is obvious. Thus, the ability to ensure the safety of railwaytransportation, orderly and efficient operation becomes very important.Since therailway rail particularity, and the locomotive speed, the accident will cause greatharm and loss to the lives of passengers and property.In this article, by collecting locomotive driver driving video signal andreal-time face detection of video signal, then carry the human eye positioning, getthe real-time analysis of the current driving state. When the driver is already infatigue or away from the work location, it alarm, to reduce fatigue caused by thedriver itself, thereby reduce or even avoide the loss of life and property duringrailway transport process.This paper analyzes the broader application of face recognition methods andalgorithms for analysis using the following methods:First, analyzes the AdaBoost algorithm used for face positioning, buildOpenCV-based face recognition database tracking system dynamic color for theframe, and traced to people face to analyze the information obtained and the humaneye positioning. It then analyzes the human eye in the video state machine algorithmhas been training through the mental state of the driver driving face, especially theeye driving habits. With a subsequent analysis of real-time picture comparison, getthe current driver’s driving state, and then determine whether it meets our standardsset by fatigue.This paper studies the color-based face detection method and AdaBoostalgorithm. From the definition of Haar shape features rectangular, describes theprocess composition of weak classifiers to strong classifier. Then through the strongclassifier cascade, consisting cascade classifier for face positioning, not nonlysaving time, also speed up the processing speed, with Camshift algorithm to achievehuman face tracking on the motion. Experimental results show that this combinationalgorithm suitable for different people and different lighting situations facerecognition. Situations with different postures of the face can also be identified.Through the experiment, expression changes and partial face was obscured, theyalso can be detected and tracked. And the success rate is high.The eye template matching method and PCA methods are used to identify andlocate the eyes, and then select the variable step length to find way to achieve thedynamic position of the eye. Among the actual tests, it can achieve a variety of abnormal conditions of the human eye positioning. The basic fatigue detection isrealized.