Online configuration of PTZ camera networks

A method for PTZ camera re-configuration oriented toward tracking applications and surveillance systems is presented. Pan, tilt, and zoom configurations are computed based on visual constraints given by a coverage model of the camera system. The visual constraints are transformed into geometric constraints by the coverage model, and the final pan, tilt, and zoom configurations are averaged by a consensus algorithm. The approach is defined in a multi-agent framework by designing a distributed algorithm that enables scalability of the system and cooperation between the cameras. Experimental results show successful camera handoff and target tracking.