摘要
行人检测技术是定点视频监控中数据采集分析的一项关键技术,实现对系统中运动行人自动检测,有着非常重要的意义。由于监控视频中远景区域中的运动行人目标相对较小,对运动行人的检测常存在漏检和误检等问题。通过设置辅助变焦摄像头,对远景区域进行数据采集,使得采集到的图像分辨率得到提高,再进行检测分类处理,从而解决远景区域中运动行人漏检率和误检率高的问题。应用基于运动特征的目标检测算法和卷积神经网络模型,设计并开发了配置辅助变焦摄像头的定点视频监控系统行人检测平台,实现了对运动行人的自动检测。实验结果表明,设计搭建的平台可有效提高对运动行人检测的准确率。
Pedestrian detection technology is a key technology for data collection and analysis in fixed-point video surveillance.It is of great significance in building the automatic system of moving pedestrians detection.Because of the reason that the moving pedestrian target in the remote area in the surveillance video is relatively small,the detection of moving pedestrians has problems of missed detection and false detection.By setting an auxiliary zoom camera to collect data in the remote area,the resolution of the collected image is improved,and then the detection and classification processing is performed,so as to solve the actual problem of high missed detection rate and the high false detection rate of moving pedestrians in the remote area.In order to realize the automatic detection of moving pedestrians,we use the key algorithm of the target detection and the convolutional neural network model,design and develop a fixed-point video surveillance pedestrian detection platform equipped with auxiliary zoom cameras.Experimental results show that the platform designed and built in this paper effectively improves the accuracy of moving pedestrian recognition.
作者
杨姝
于闯
蒋宁
YANG Shu;YU Chuang;JIANG Ning(College of Educational Technology, Shenyang Normal University, Shenyang 110034, China;Computer and Basic Mathematics Education Department, Shenyang Normal University, Shenyang 110034, China;School of Computer, Shenyang Aerospace University, Shenyang 110136, China)
出处
《沈阳师范大学学报（自然科学版）》
CAS
2021年第4期348-352,共5页
Journal of Shenyang Normal University:Natural Science Edition
基金
辽宁省教育厅科学研究经费项目(JYT2020112)。
关键词
定点
视频监控
行人检测
fixed-point
video surveillance
pedestrian detection
作者简介
杨姝(1963-),女,辽宁沈阳人,沈阳师范大学教授,博士;通讯作者:于闯(1977-),男,辽宁凤城人,沈阳师范大学实验师,硕士。
