Practical Dynamic Logic GMTI Tracker

Dynamic logic (DL) method provides an efficient algorithm for tracking targets in clutter. The DL tracker overcomes the combinatorial complexity of tracking by substituting the discrete assignment problem with a problem of continuous optimization of the maximum likelihood criterion in a low-dimensional parameter space. DL enables efficient optimization of the likelihood via a combination of expectation-maximization optimization of a smoothed likelihood with gradual reduction of the likelihood smoothing. The non-combinatorial complexity enables the DL tracker to process a large number of data frames in a single optimization process, resulting in high detection performance of weak targets. In this paper, we experimentally evaluate practical usefulness of DL for ground moving target indicator (GMTI) target tracking. We evaluate quality of DL track estimates by constructing receiver operating characteristics (ROC) curves for a simulated scenario. We also evaluate performance of the DL tracker incorporating measurements from moving platforms and approximate road priors.