Research On Driver Fatigue Driving Detection System

With the sustained and rapid economic development,China's road transport developed more and more convenient,but also brought a series of problems:A sharp rise in the number of vehicles,unbearable traffic jams,frequent traffic accidents,even a sharp increase in malignant traffic accidents in recent years,resulting in spurting rise in the number of casualties and our country's serious economic losses and adverse social impact.Fatigue driving has become the main reason for frequent traffic accidents,how to effectively prevent the occurrence of fatigue driving has become an urgent problem,which reduce traffic accidents is very important.In the article,CCD camera installed in the cab real-time shooting driver's face, the video image is preprocessed from the CCD camera,then the preprocessed gray image is used to detect face by AdaBoost algorithm based on Haar feature,then the detected face region is used to detect eye by eye detection algorithm based on gray integral projection,Canny operator is used to detect the human eye edge,so that the contour of the human eye could be revealed in the grayscale image.Through distance calculation of the upper and lower eyelids in the eye contour to determine the degree of opening about the driver's eyes. According P80 guidelines of PERCLOS algorithm, combined with the open level of the driver's eyes to determine whether the driver is a state of fatigue.This process is the basic detection process of fatigue detection system.AdaBoost algorithm based on Haar features and eyes detection algorithm based on gray integral projection in the System is the core algorithm of this article.In this paper the basic principles and implementation process of these two algorithms were introduced,and through a large number of samples to verify these two algorithms.According to the analysis of the verified results, the detection rate of AdaBoost face detection algorithm based on Haar feature and the human eye detection algorithm based on gray integral projection was relatively high,which detection speed was relatively fast,through the analysis of algorithm insufficient to provide improved direction.And comparing the proposed new eye detection algorithm with the original eye detection algorithm,the proposed new eye detection algorithm is more suitable for fatigue detection system in this article.PERCLOS algorithm in this article is the best driver fatigue algorithm.According to this algorithm,fatigue detection system could real-time detect whether the driver is a state of fatigue during driving.