A mean track approach applied to the multidimensional assignment problem

The primary contribution of this paper is the introduction of a new preprocessing method to eliminate unlikely observation-to-track pairs with medium size target-measurement spacings before filter gating is carried out. The proposed method follows a track-oriented approach, with the different track hypotheses being contained in target trees. A mean track is computed for each target tree to capture as much variability as possible from the track hypotheses which are in the tree. Each newly received measurement is first gated with the mean track; only if the outcome of the test is positive, the measurement is gated with all the track hypotheses in the tree. The experimental results support the theoretical claims in the paper and show that a significant reduction in the number of observation-to-tracking pairing tests is achieved.