Variable Gating MLPDA for early-extraction of a target track in high false density environments

摘要

MLPDA (Maximum Likelihood Probabilistic Data Association) is an effective method to extract a target track in high false density environments, but it requires very high computational load. To resolve the high computational burden, we propose VG-MLPDA (Variable Gating MLPDA) which consists of two steps: rough search and detailed search. At the rough search step, it finds the maximum likelihood track among search points with large gate size, thus the number of search points can be smaller than conventional MLPDA. Then, at the detailed search step, it maximizes the track likelihood while reducing the gate volume. Simulation results show that our proposed VG-MLPDA can extract the target track earlier than the conventional track-oriented MHT (Multiple Hypothesis Tracking).