基于PERCLOS的疲劳驾驶检测研究  

Research on Fatigue Driving Detection Based on PERCLOS

作　　者：刘东华 刘庆华[1] 李杨[1] 黄凯枫 LIU Donghua;LIU Qinghua;LI Yang;HUANG Kaifeng(Jiangsu University of Science and Technology,Zhenjiang 212003)

机构地区：[1]江苏科技大学,镇江212003

出　　处：《计算机与数字工程》2022年第7期1583-1586,1608,共5页Computer & Digital Engineering

摘　　要：疲劳驾驶成为了导致交通事故发生的重要因素之一,并呈现出逐年递增的趋势,若能设计出一种在驾驶员发生疲劳时,就能检测出驾驶员处于疲劳状态对其进行警告,这样就可以较好地降低交通事故的发生的概率。论文介绍了一种检测驾驶员在驾驶过程中是否为疲劳驾驶的方法,基于PERCLOS对驾驶员进行疲劳检测。首先对设备采集的人脸图像进行肤色分割,根据程序中设定的肤色阈值,确定图像中属于肤色的区域,对得到的肤色区域进行眼睛追踪,人脸特征部位进行积分投影操作,获取眼睛区域,捕捉眼睛的实时状态,并对眼睛的面积进行计算,最后根据疲劳值来判定疲劳状态。根据实验证明,该方法满足车载、实时、非接触的基本要求,并能准确地做出对驾驶员驾驶过程中是否出现疲劳的判定。Fatigue driving has become one of the important factors leading to traffic accidents,and it is increasing year by year.If a warning method can be designed when the driver is tired,the probability of traffic accidents can be reduced.This paper in⁃troduces a method to detect whether the driver is tired driving in the driving process,and based on PERCLOS,the driver's fatigue is detected.Firstly,the skin color of the face image collected by the device is segmented.According to the skin color threshold set in the program,the skin color and non skin color areas in the image are determined.The area of skin color obtained is located by the eyes.The integral projection is carried out from the feature part of the face.After positioning the eyes,the real-time state of the eyes is tracked,and to calculate the area of the eyes when they are open and closed.Finally,the fatigue is determined according to PER⁃CLOS fatigue criterion.According to the result indicate that the method can satisfy the basic requirements of real-time and non-con⁃tact,and can accurately judge whether the driver is tired during driving.

关 键 词：疲劳驾驶 肤色分割 积分投影 眼睛定位 PERCLOS 

分 类 号：TP315[自动化与计算机技术—计算机软件与理论]