Parameter Tuning for Maritime Track-Before-Detect

Track-before-detect (TBD) is a joint detection and tracking approach that takes advantage of a targets motion over time. To work effectively, it requires latent parameters to describe the underlying dynamics of the targets relative to the sensor system. Determining the latent parameters is critical for success, with errors potentially resulting in failure of the tracking algorithm. In this paper, we propose a parameter tuning method for the Bayesian TBD algorithm known as the Bernoulli particle filter. The parameter estimation formulation uses a Bayesian framework with the particle Markov-chain Monte Carlo algorithm. Experimental results for a maritime radar scenario are used to validate the new algorithm.