Amplitude Information Aided Robust Multi-Bernoulli Filter for Marine Target Tracking

The information of clutter rate and detection probability is very important for the Bayesian multi-target filters based on random finite sets (RFS). However this information is difficult to learn on line in marine target detection applications. The robust multi-Bernoulli filter (RMB) can accommodate the unknown clutter rate and detection probability, thus it is a rational alternative in this challenging situation. But this method only exploits the kinematic information when calculating the measurement likelihood, therefore its performance is not ideal if the targets and clutter are spatially close. In this paper, the amplitude information (AI) of the target and sea clutter is incorporated into the RMB filter, which helps to distinguish targets from clutter better, and further gives an improved performance in the estimation of target state, cardinality, as well as clutter rate. The performance of the proposed algorithm are evaluated via tracking experiments for multiple fluctuating targets of Swerling type 1 in heavy tailed K distributed sea clutter.