基于头部姿态眼睛差分定位的驾驶员疲劳检测  

Driver Fatigue Detection Based on Head Pose for Eye Differential Positioning

作　　者：唐阳山[1] 徐忠帅 杨语尧 TANG Yang-shan;XU Zhong-shuai;YANG Yu-yao(College of Automobile and Transportation,Liaoning University of Technology,Jinzhou 121000,China)

机构地区：[1]辽宁工业大学汽车与交通学院,锦州121000

出　　处：《科学技术与工程》2018年第31期230-234,共5页Science Technology and Engineering

基　　金：辽宁省科技厅联合基金(201602375)资助

摘　　要：基于面部特征的驾驶员疲劳判断是应用最广泛的方法,而眼睛睁闭程度最直接表达驾驶员的精神状态。传统的眼睛检测方法受环境、传输、头部姿态的影响,眼睛的定位精度不高,从而导致疲劳分析不准确。提出一种基于脸部图像灰度差进行眼睛检测的方法:正常情况下在人脸上半部只有眼睛进行睁闭的活动,故眼部区域灰度会发生变化,由此来进行标定。该方法主要包括基于Ada Boost算法的人脸识别、图像预处理、眼睛的检测、积分投影法计算眼睛的高宽比,以及基于PERCLOS(percentage of eyelid closure)准则的驾驶员疲劳判断。最后分别基于头部左转、右转和正视三种情况下进行实验。根据结果表明该方法能够较好地进行眼睛的检测,对于进行驾驶员的疲劳判断有极大的意义。The driver fatigue judgment based on facial features is the most widely used method,and the degree of eye closure is the most direct expression of the driver s mental state.The traditional eye detection method is affected by environment,transmission and head posture,and the positioning accuracy of eyes is not high,resulting in inaccurate fatigue analysis.A method of eye detection based on the gray level difference of the face image is proposed.In normal cases,only the eyes in the half of the face are open and closed,so the gray level of the eye area will change,which can be calibrated.The method mainly includes face recognition based on AdaBoost algorithm,image preprocessing,eye detection,integral projection method to calculate the height width ratio of eyes and driver fatigue judgment based on PERCLOS criterion.Finally,the experiment is carried out in three cases,based on the left turn of the head,the right turn and the face.The results show that the method can be used to detect the eyes well,and it is of great significance to the fatigue judgment of the driver.

关 键 词：灰度变化 人脸识别 眼睛检测 积分投影法 PERCLOS准则 

分 类 号：U491.255[交通运输工程—交通运输规划与管理]