A data association approach based on the MHSDA algorithm for passive optical sensors with partially overlapped field of view

This paper focus on solving the data association problem for passive optical sensors with partially overlapped field of view (FoV). Utilizing multiple sensors can expand the monitoring range of surveillance systems. In order to ensure completely coverage the monitoring area, the FoV of adjacent sensors should partially overlapped. Passive optical sensors can obtain angle-only measurements of targets and cannot intuitively determine which targets are located within the overlapped FoV. In this condition, due to factors such as “ghost” and targets number unknown, the data association accuracy of classical algorithms are relatively low. The recently proposed multiple hypothesis S-dimensional assignment (MHSDA) algorithm maintains the ability of the assignment algorithm to deal with "ghost" and counteract the effect of target number unknown through multiple hypothesis approach. In this paper, an experimental environment based on CCD cameras is established and the the effectiveness of the MHSDA algorithm is validated under the condition that the FoV of sensors are partially overlapped, laying an important foundation for the practical application of the MHSDA algorithm for multi-sensor surveillance systems.