Multiobject Tracking with Track Continuity: An Efficient Random Finite Set Based Algorithm

Abstract

We propose a random finite set (RFS) based algorithm for tracking multiple objects while maintaining track continuity. In our approach, the object states are modeled by a combination of a labeled multi-Bernoulli (LMB) RFS and a Poisson RFS. Low complexity is achieved through several judiciously chosen approximations in the update step. In particular, the computationally less demanding Poisson part of our algorithm is used to track potential objects whose existence is highly uncertain. A new labeled Bernoulli component is generated only if there is sufficient evidence of object existence, and then the corresponding object state is tracked by the more accurate but more complex LMB part of the algorithm. Simulation results for a challenging scenario demonstrate an attractive accuracy-complexity tradeoff and a significant complexity reduction relative to other RFS-based algorithms with comparable performance.