Improvement in track-to-track association from using an adaptive threshold

This paper considers the performance of two track-to-track association algorithms. The first bases association decisions on a chi-squared distance between tracks and a fixed significance threshold to determine when no association is allowed. The second algorithm finds the maximum a posteriori probability (MAP) set of associations between tracks from two independent tracking systems. For tracks whose state estimates are characterized by Gaussian distributions, the second algorithm may be viewed as a version of the first algorithm with an adaptive threshold. This paper examines the performance of these two algorithms in terms of expected fraction of correct matches of tracks from system 1 to tracks from system 2 and finds that the adaptive threshold algorithm performs as well or better than the fixed threshold algorithm and that adjustments to the adaptive threshold generally produce little or no benefit.