Association gate-based GM-PHD for high-efficiency tracking

With measurement-associated intensity update, a novel scheme of high-efficient tracking of extended target under the frame of Gaussian mixture probability hypothesis density (GM-PHD) filtering is presented. Accordingly to the extended target GM-PHD filtering, the persistent and newborn targets are distinguished, but their intensities are updated using the entire set of observation, its inefficiency is explicit, that is, the association between the measurement set and the target set is not considered. We propose the measurement association gate for the pre-selection of observations set, so as to get rid of the disassociated measurements set and the useless computation is reduced reasonably. Meanwhile the associate gate is also employed for measurements partitioning for the case of extended target. The simulation results are discussed and the improved efficiency of proposed method is confirmed.