Comparative evaluation of two Bayesian approaches to multiple target tracking

A comparative performance evaluation of two different Bayesian approaches to multiple target tracking is presented in this work. The first approach, namely the "Joint Integrated Probabilistic Data Association" filter, (JIPDA) outlined in the work Musicki and Evans (2002), will be compared to the Multiple Hypothesis Tracker (MHT), introduced in the work of Bar-Shalon and Blair (2000). The JIPDA is an extension of the IPDA, introduced in in the work Musicki and Evans (1994), to the multiple-target case. The Multiple Hypothesis technique is applicable to both the single- and multiple-target scenarios, provided that the necessary algorithmic provisions are made. A standard Kalman filter is used at the core of both approaches, to deal with the estimation of the states of tracks. In particular, these processing techniques will be evaluated in their application to data from a marine surface RADAR system, taking into consideration the measurement origin uncertainty, i.e. if any of the considered measurements has been originated by an actual target in the scenario of interest, or it is a false measurement, having its origin in a different physical (from the observed environment) or electronic (at sensor level) phenomena.