Efficient Adaptive Detection Threshold Optimization for Tracking Maneuvering Targets in Clutter

In this paper, we focus on the adaptive prior detection threshold setting problem to optimize the overall performance of the joint detection-tracking system for maneuvering target tracking in clutter. It is shown that our problem can be reduced to the information reduction factor (IRF) maximization by Gaussian fitting of maneuvering target Markovian switching dynamics via moment matching, even for the case with the nonlinear measurement equation. Our proposed adaptive threshold setting method outperforms the conventional threshold setting approaches greatly and also exhibits a mildly improvement in comparison with the earlier method for this problem in terms of tracking performance, especially in track loss percentage (TLP). However the computational burden of our method is reduced significantly because in our method generally only one IRF corresponding to the common validation region, not the every IRF corresponding to the individual model-conditioned validation region, is needed for threshold optimization at each time step and an approximate closed-form solution can also be obtained for the special case of the Neyman-Pearson (NP) detector.