Track maintenance using measurements of target extent

A simple elliptical target model is proposed for exploiting sensor measurements of target extent within tracking filters. This extra information can provide a most useful aid to track retention for closely spaced moving targets with measurement association uncertainty. Tracking performance has been demonstrated for an extended Kalman filter (EKF) implementation using second-order (constant velocity) and co-ordinated turn dynamics models with nearest neighbour and joint probabilistic data association measurement selection schemes. Under certain conditions, this EKF implementation is prone to divergence and it is shown that a straightforward particle filter can avoid this problem.