Localization of distributed wireless cameras

Abstract

Cooperative cameras enable monitoring wide areas and detecting actions and events on a large scale. Due to hardware advancements and economic factors, distributed networks are becoming widely used for a variety of applications ranging from traffic monitoring and surveillance in shopping malls to sports coverage. However, the localization of a large number of cameras in a wide area is not a trivial task. Manual methods are time consuming and can be inaccurate over time. For this reason, we propose an algorithm that uses measurements from the observed objects to perform pair-wise automatic localization of a distributed set of cameras with non-overlapping fields of view. We use the temporal information derived from trajectory information to estimate unobserved trajectory segments, which are then used to estimate the position of the cameras on a common ground plane. Furthermore, the exit-entrance direction of the moving objects is used to estimate the relative orientation of adjacent cameras. We demonstrate the algorithm on a distributed network of simulated cameras with wireless communication and compare it with centralized state-of-the-art approaches.