GM-CPHD and ML-PDA applied to the Metron multi-static sonar dataset
Rodica Georgescu, Peter Willett, Steven Schoenecker · 2010
The Gaussian Mixture Cardinalized Probability Hypothesis Density (GM-CPHD) Tracker and the Maximum Likelihood-Probabilistic Data Association (ML-PDA) Tracker were applied to the Metron simulated multi-static sonar dataset created for the MSTWG (Multistatic Tracking Working Group). The large number of measurements at each scan was a problem for the GM-CPHD. Winnowing by a detection test on SNR and Doppler followed by predetection fusion (contact sifting) had to be performed before tracking with the GM-CPHD to obtain good performance. Due to the nature of its likelihood formulation, ML-PDA was able to process the measurements directly and reasonable results for all five scenarios were achieved. Plots of the tracks obtained on the five scenarios the Metron dataset with both trackers are provided.