基于人脸特征的疲劳检测在矿山远程驾驶中的应用  

Application of Fatigue Detection Based on Face Feature Classification in Mine Remote Driving

作　　者：刘旭 战凯[1,2,3] 于骞翔 张元生[1,3,4] LIU Xu;ZHAN Kai;YU Qianxiang;ZHANG Yuansheng(BGRIMM Technology Group,Beijing 100160,China;University of Science and Technology Beijing,Beijing 100083,China;Beijing Key Laboratory of Nonferrous Intelligent Mining Technology,Beijing 102628,China;BGRIMM Intelligent Technology Co.Ltd.,Beijing 102628,China)

机构地区：[1]矿冶科技集团有限公司,北京100160 [2]北京科技大学,北京100083 [3]金属矿山智能开采技术北京市重点实验室,北京102628 [4]北京北矿智能科技有限公司,北京102628

出　　处：《有色金属工程》2022年第4期124-130,共7页Nonferrous Metals Engineering

基　　金：山东省重大科技创新工程项目(2019SDZY05)。

摘　　要：针对智能矿山井下运输设备远程操作过程中驾驶员疲劳作业的安全隐患问题,设计了一种驾驶员人脸特征的疲劳检测系统。首先规定不同行为的疲劳驾驶等级,其次对相机采集到的视频信息进行预处理,采集不同角度下驾驶员面部图像样本,采用ASM算法对人脸建模,计算驾驶员的面部特征点定位、眼睛闭合频率、打哈欠频率等面部信息,并依靠深度学习神经网络的自适应特征提取进行数据分析,判断驾驶员疲劳程度。自建人脸数据库和安徽某矿山现场测试表明,该系统能够对驾驶人员在高作业强度、多设备协同下产生的违规操作起到预警和防止的作用。Aiming at the safety hidden danger of driver fatigue operation in the remote operation of underground transportation equipment in intelligent mine,design a fatigue detection system based on driver face features.First,specify the fatigue driving level of different behaviors.Second,preprocess the video information collected by the camera,collect the driver′s face image samples under different angles,model the face with ASM algorithm,and calculate the driver′s facial information such as facial feature point location,eye closure frequency and yawning frequency.Based on the adaptive feature extraction of deep learning neural network,the data are analyzed to judge the driver′s fatigue degree.The face database and the field test of a mine in Anhui province show that the system can warn and prevent the illegal operation of drivers under high operation intensity and multi equipment cooperation.

关 键 词：人脸特征 深度学习 疲劳驾驶 

分 类 号：TD67[矿业工程—矿山机电]