基于云平台和CNN的交通视频监控车辆检测方法及应用  

Detection Method and Application of Traffic Video Monitoring Vehicles Based on Cloud Platform and CNN

作　　者：王芳彬 WANG Fangbin(Nanping Fuyin Expressway Co.,Ltd.,Nanping,Fujian 353000,China)

机构地区：[1]南平福银高速公路有限责任公司,福建南平353000

出　　处：《自动化应用》2023年第24期227-228,231,共3页Automation Application

摘　　要：为提升交通视频监控车辆检测的检测跟踪准确率,本文提出基于云平台和卷积神经网络(CNN)的交通视频监控车辆检测方法。采集车辆运动轨迹,并预处理车体较高或车辆目标中心点不在实际车道内的车辆运行轨迹;搭建交通视频监控云平台,提取运动车辆特征;利用CNN设计交通视频监控车辆目标检测算法。结果表明,本文方法的车辆检测准确率在98%以上,车辆检测结果与车辆行驶的实际情况更加接近,检测跟踪更加精准。To improve the detection and tracking accuracy of traffic video surveillance vehicle detection,this paper proposes a traffic video surveillance vehicle detection method based on cloud platform and convolutional neural network(CNN).Collect vehicle motion trajectories and preprocess vehicle motion trajectories with higher body height or target center points outside the actual lane.Build a traffic video monitoring cloud platform to extract features of moving vehicles.Design a vehicle target detection algorithm for traffic video surveillance using CNN.Results show that the vehicle detection accuracy of the proposed method is over 98%,and the vehicle detection results are closer to the actual driving situation of the vehicle,resulting in more accurate detection and tracking.

关 键 词：云平台 卷积神经网络 监控 车辆 

分 类 号：U492[交通运输工程—交通运输规划与管理]