Track decoupling: linear joint IPDA (LJIPDA) and multi-target linear IPDA (MLIPDA)

This paper presents a new approach to multi-target tracking. Rather than forming complex hypotheses based on all possible combinations of measurement origins, we attempt to decouple individual tracks based on the probabilities of measurement origins. Two such algorithms, both based on the IPDA algorithm, are presented in this paper. One, which we call linear joint IPDA (LJIPDA), recalculates IPDA using the probabilities of measurement origin. The other, which we call multitarget linear IPDA (MLIPDA), uses the probabilities of measurement origin to modify IPDA results. Both algorithms are recursive and yield formulae for both data association and probability of track existence. Simulations were carried out to compare these algorithms with IPDA in a dense and non-homogenous clutter situation.