黄志超 赵红梅 陈奕迅 汤耀森



摘要:为了有效降低驾驶员疲劳驾驶、短时间未看道路等危险驾驶的概率,设计了以面部特征为研究对象的人脸识别的检测系统。该系统通过对面部识别以及眼睛和嘴的定位跟踪,提取面部疲劳信息,将跟踪的图像信息传输到 OpenCV视觉库进行图像处理,通过一系列的算法对驾驶员疲劳、危险驾驶行为进行判断、疲劳值计算,最后通过安全警报模块对驾驶员给予警报提示。经实验测试,该系统满足车载、实时性、准确性的要求,能够快速的检测驾驶员的危险驾驶状态,具有一定的实用性。
关键词:面部识别;疲劳驾驶;危险驾驶
中图分类号:TP274 文献标志码:A 文章编号:1009-9492(2021)12-0143-04
Driver Fatigue and Dangerous Driving Detection System Based on Facial Recognition
Huang Zhichao ,Zhao Hongmei ,Chen Yixun ,Tang Yaosen
(Huali College, Guangdong University of Technology, Guangzhou 511325, China)
Abstract: In order to effectively reduce the probability of dangerous driving such as driver fatigue driving and not looking at the road for a short time, a face recognition detection system based on facial features was designed. The system extracts facial fatigue information through face recognition and positioning and tracking of eyes and mouth, transmits the tracked image information to OpenCV visual library for image processing, judges the driver′s fatigue and dangerous driving behavior and calculates the fatigue value through a series of algorithms, and finally gives an alarm to the driver through the safety alarm module. The experimental test shows that the system meets the requirements of on-board, real-time and accuracy, can quickly detect the dangerous driving state of drivers, and has certain practicability.
Key words: facial recognition; fatigue driving; dangerous driving
0 引言
根据中国质量新闻网数据显示,2020年中国汽车保有量已达2.81亿辆,驾驶人员达到4.18亿人。随着我国汽车保有量的持续增加,驾驶员人群的日益庞大,交通事故频发的风险随之增高,而疲劳驾驶和看手机、短时间未看道路等危险驾驶行为是交通事故发生的主要原因。
面部识别技术从20世纪60年代的早期研究阶段,到现在的深度学习的突破阶段,发生了巨大的转变,广泛应用于各个领域。基于面部识别的疲劳驾驶检测方法,因为实时性好,准确度高等特点,受到了广泛的关注。本文将面部识别技术和驾驶员疲劳检测融合起来,在汽车环境中,搭建以面部识别为核心的系统硬件平台。通过对驾驶员面部的扫描,找到人脸的位置并跟踪,提取眼睛和嘴……