基于双目深度图像的自动扶梯乘客危险行为识别与预警系统  

A Passenger Dangerous Behavior Recognition and Early Warning System of Escalator Based on RGB-D Sensor

作　　者：欧阳惠卿[1] 舒文华[1] 李行 李杨[3] OUYANG Huiqing;SHU Wenhua;LI Xing;LI Yang

机构地区：[1]上海市特种设备监督检验技术研究院,上海200062 [2]南京美基森信息技术有限公司,南京210093 [3]南京大学,南京210093

出　　处：《中国电梯》2020年第14期36-39,42,共5页China Elevator

基　　金：国家市场监督管理总局科技项目(2019MK019)。

摘　　要：基于双目深度视觉传感器,设计了一种融合双目立体匹配算法和深度神经网络模型的方法,针对乘客危险行为识别与预警的需求,构建了一套可应用于自动扶梯场景的乘客危险行为识别与预警系统。实验结果表明,该系统能有效识别乘客摔倒、逆行、探头、出入口滞留以及携带婴儿车和大件行李等多种危险行为,并能自动预警。Based on the binocular depth(RGB-D)sensor,a method of integrating binocular stereo matching algorithm and depth neural network model is designed.Aiming at the needs of passenger dangerous behavior recognition and early warning,a set of passenger dangerous behavior recognition and early warning system which can be applied to escalator scene is constructed.The experimental results show that the system can effectively identify various dangerous behaviors,such as passenger falling,retrograde,probe,entrance and exit detention,carrying pram and heavy luggage,and can automatically warn.New ideas and solutions for improving the operation safety of escalators are provided.

关 键 词：自动扶梯 行为识别 双目深度 深度学习 人工智能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程] TH236[机械工程—机械制造及自动化]