Tracking performance under time delay and asynchronicity in distributed camera systems

Distributed camera systems are typically used to simultaneously track multiple targets. Communication between cameras enables the ability to monitor targets in a complex environment where occlusion happens. Asynchronicity and time delay are among the most important factors that affect tracking error. We provide a feedback law controlling the pan and tilt of each camera to track the centroid of a cluster of point targets in the field of view. Assuming that the estimates for target motion are available only infrequently and asynchronously when occlusion happens, we compute a worst-case lower bound on the frequency for exchanging estimates between cameras. Our results may help camera system designers to determine the response time and tracking ability for distributed camera systems.