摘要
为实现人群密集场所客流安全隐患早发现,辅助管理人员早决策,人群聚集风险区早疏散,提升对灾难的预见性和主动性。在国内外人群异常聚集监测预警现状分析基础上,对比分析得出监控视频分析技术是解决人群密集场所精准预警难题较为理想的解决方案;构建以视频智能分析的人群计数、密度估计、行人追踪、活动烈度识别为核心技术的人群密集场所风险预警技术框架;将该技术框架应用到某大型商圈的商业街区,获得监控区域内的人群总数、密度分布、行人轨迹和异常活动等特征。结果表明:提出的基于视频分析的人群密集场所风险预警技术框架可为城市大型商圈、交通枢纽、大型活动场所等城市公共场所的安全管理提供参考和借鉴。
In order to realize the detection on the potential safety hazard of passenger flow in the crowded places early,assist the managers to make decision early,evacuate the crowd in the gathering risk area early,and improve the foresight and initiative on the disasters,based on the analysis on the current situation of monitoring and early-warning on the abnormal gathering of crowd at home and abroad,it was concluded that the analysis technology of monitoring video was the more ideal solution to solve the problem of accurate early-warning in the crowded places through the comparative analysis.A technical framework of risk early-warning for the crowded places was constructed,which took the crowd counting,density estimation,pedestrian tracking and activity intensity recognition as the core technologies,and it was applied in the commercial block of a large business district,then the characteristics of the total number of people,density distribution,pedestrian trajectory and abnormal activities in the monitored area were obtained.The results showed that the technical framework of risk early-warning for the crowded places based on the video analysis can provide reference for the safety management of urban public places such as large commercial districts,transportation hubs and large event venues.
作者
陈冲
白硕
黄丽达
王晓萌
刘春慧
CHEN Chong;BAI Shuo;HUANG Lida;WANG Xiaomeng;LIU Chunhui(Department of Engineering Physics,Tsinghua University,Beijing 100084,China;Beijing Global Safety Technology Inc.,Beijing 100085,China)
出处
《中国安全生产科学技术》
CAS
CSCD
北大核心
2020年第4期143-148,共6页
Journal of Safety Science and Technology
基金
北京市科委项目(Z181100009018010)
中国博士后科学基金项目(2019M660661)。
关键词
人群密集场所
人群计数
人群密度
异常行为
监控预警
crowded places
crowd counting
crowd density
abnormal behavior
monitoring and early-warning
