Distributed partitioning strategies with visual optimization for camera network perimeter patrolling

This work addresses the boundary patrolling problem, where a smart camera network undertakes the task of monitoring the perimeter of an environment so as to detect anomalies and track possible intrusions. Here, a distributed solution is sought based on the definition of a suitable functional that accounts both for the equitable partitioning of the available space and for the quality of vision of the patrolled area, and admits a unique optimal solution. The optimization of such functional leads to the design of an algorithm relying on a symmetric gossip communication protocol among the neighboring cameras. The theoretical results prove the correctness of the approach and the numerical simulations on a realistic scenario confirm its validity.