An improved Multitarget Multi-Bernoulli filter with cardinality corrected

The Multitarget Multi-Bernoulli (MeMBer) filter has been proposed as a computationally tractable approximation of the multitarget Bayes filter, but with significant bias in the cardinality estimates. In this paper, a new improved MeMBer filter is proposed as a solution to address these limitations, with stable and accurate cardinality estimates achieved. The biases derived from the two approximations used in the derivation of the MeMBer filter corrector are analyzed. The proposed filter gives a new form of the updated tracks in the filter corrector, and the over-estimated cardinality in dense clutter environment is corrected. The Gaussian Mixture (GM) approximation is used to implement the proposed filter. Simulation results demonstrate the effectiveness of the proposed algorithm.