Vessel Tracking Under Random Finite Set Framework

The developments of vessel tracking systems have been significantly improved in recent years. A large number of approaches have been investigated, for vessel tracking in various environments. However, data association is still a challenge. As the number of clutter increasing, measurements which originated from vessels could not be easily classified at each step. Hence the filter could not keep the robust during the estimation. The PHD (Probability Hypothesis Density) filter is therefore presented for vessel tracking in such environments, which does not require an enumeration of measurement-to-target association during the filtering process. The key idea is to consider both states and measurements as set-valued state and set-valued measurement, respectively. Hence the data association issue is avoided in Bayesian framework. A comparative study based on simulations demonstrates the feasibility and the reliability of the proposed approach in 2D Cartesian coordinates.