Publication

Abstract : An efficient approach to camera deployment enhances the cost-effectiveness and functionality of a multi-camera surveillance network. Traditional camera placement strategies involve maximizing total coverage over a surveillance region with predefined camera locations by optimizing the camera orientations. However, some modern approaches allow users to specify the desired total coverage over the surveillance region, and the algorithms subsequently determine the optimal number of cameras, as well as their exact locations and orientations, to meet this specified coverage. The Reward Penalty Score (RPS) algorithm and Extended Greedy Grid Voting (EGGV) algorithm are two innovative algorithms designed to attain the coverage constraint by proposing an optimal number of cameras and their locations and orientations. In certain scenarios, the number of cameras available for designing the surveillance network is limited, which should be treated as an input constraint rather than a specified coverage requirement. Under such conditions, the primary objective becomes maximizing total coverage with the given number of cameras by determining their optimal locations and orientations. Currently, there are no widely recognized algorithms specifically designed to handle this particular scenario. In this paper, we effectively modify the RPS and EGGV algorithms (m-EGGV and m-RPS), adapting them to optimally deploy the specified number of cameras over the entire surveillance region in an efficient manner to maximize coverage. Additionally, by employing the m-EGGV and m-RPS algorithms, we address the scenario of coverage loss resulting from the failure of one or more cameras. These modified algorithms facilitate the relocation of a subset of potential cameras which can alleviate this loss in coverage caused by the failure of cameras. The m-EGGV and m-RPS algorithms demonstrate a robust performance through extensive testing in diverse simulation environments.