摘要
论文构建了基于多路摄像机、微光夜视仪、红外热像仪的音频、视频等多源信息集成和融合的行为识别体系结构。提出了大型客运站的异常行为定义及识别流程,通过构建个体简单行为模型,并计算个体简单行为概率,推断复杂群体行为的发生概率。研究了客运站异常行为的界定和报警阀值设置,实现了客运站异常行为的有效识别和及时报警,为智能视频监控技术提供新的理论和方法。
An architecture of human behavior identification based on multi-information integration and fusion of multi-camera,low light level night vision,infra-red thermal audio and video imaging was constructed,and abnormal behavior definition and identification processes for large-scale passenger transport station was put forward.Through building an individual simple behavior model and calculating its behavior probability,the group complex behavior probability and identification were obtained.Then,by studying the abnormal behavior definition and alarm threshold setting,effective abnormal behavior identification and timely warning were realized.The study can provide a new theory and method for intelligent video surveillance technology.
出处
《公路交通科技》
CAS
CSCD
北大核心
2009年第S1期58-61,87,共5页
Journal of Highway and Transportation Research and Development
关键词
智能运输系统
信息融合
异常行为识别
特征分析
ITS
multi-information fusion
abnormal behavior identification
character analysis
