Robust Real-Time Tracking for Visual Surveillance

摘要: This paper describes a real-time multi-camera surveillance system that can be applied to range of application domains. integrated is designed observe crowded scenes and has mechanisms improve tracking objects are in close proximity. The four component modules described this (i) motion detection using layered background model, (ii) object based on local appearance, (iii) hierarchical recognition, (iv) fused multisensor multiple features geometric constraints. approach complex scene validated against number representative real-world scenarios show robust, analysis performed.