中文摘要：

针对当前公共安全监控系统在应对突发事件时,因监视器数量多、信息量大、事件变化快等因素而导致的不能及时有效地对移动对象进行跟踪的问题,提出了一种基于拓扑结构的多摄像头移动对象辅助协同跟踪方法。首先将城市路网中的摄像头监控区域抽象为虚拟节点定位至结点弧段模型中,构建摄像头网络拓扑结构;根据拓扑关系搜索出与发现移动对象的摄像头关联的摄像头集合;再分析摄像头间的空间邻近关系和时间差异,获取时序关系最优的摄像头组作为重点监控区域进行移动对象跟踪,并定期进行协同更新。最后建立了一个模拟监控系统验证了利用多摄像头协同方法实现对移动对象实时跟踪的有效性。

英文摘要：

Due to the factors of large amount of monitors and data,event changes quickly and so on,security surveillance system can′t track moving objects timely and effectively in response to emergencies.The paper presents a coordinating multiple cameras approach to assist tracking moving objects based on network topological structure in urban environment.Firstly,the fields of view of cameras were abstracted as points located in node-arc model with linear referencing system to construct a network topological structure between cameras and urban road.According to the network topological structure,we search some cameras which have spatial neighborhood relation with first camera,and analyzing proximity relation of cameras and to find out the time difference between cameras.Then we set some cameras as the main monitoring group for tracking moving objects.As the time goes on,we could update the monitoring group by replacing some cameras in which moving objects can＇t appear according to the queue principle.In order to verify the validity of the cooperative multi-camera approach,we establish a system based on traffic simulation model that simulates the real traffic circumstance,and randomly select some moving objects that generated by the model to have a test.The experiment result demonstrates that the method is able to provide directive information consisted of map and surveillance information which can assist monitors to track moving targets successfully and effectively.