Real-time reconfiguration of PTZ camera networks using motion field entropy and visual coverage

In this paper we propose a novel dynamic camera reconfiguration algorithm for multi-camera networks. The algorithm relies on the analysis of the entropy of the acquired scene. Based on the obtained values it assigns the cameras to act as global or target sensors. While global sensors aim at guaranteeing the overall coverage of the observed space, target sensors focus on moving objects. To this aim, the entropy information is used differently according to the operation mode of each camera. The visual entropy of the captured image is used to control the configuration of the global sensors, whereas the disorder of the motion field is adopted as a metric to drive the reconfiguration of the target sensors. In order to achieve real-time operational capabilities, the algorithm works in the compressed video domain, rather than the traditional pixel domain. The algorithm has been tested in vitro, and also by deploying a camera network in a real indoor surveillance environment. In particular, re...