Generalized Labeled Multi-Bernoulli Filter Using Spectral Matching

To reduce the cardinality bias in multi-target tracking, a novel multi-Bernoulli filter called spectral matching based generalized labeled multi-Bernoulli filter is derived. By introducing the spectral information of target in update step, the proposed filter can improve tracking performance. Under the same assumption as the GLMB and CPHD recursion, the proposed recursion is unbiased in cardinality estimation. In addition, a sequential Monte Carlo implementation is proposed. Simulation results show that the proposed filter has better performance in target state and number estimation than GLMB and CPHD filter.