Improved MM-CBMeMBer Filter for Maneuvering Multi-target Tracking

摘要

To solve the problems of unstable state estimation and long running time of existing multi-model cardinality-balanced multi-target multi-Bernoulli (MM-CBMeMBer) filters, we present an improved MM-CBMeMBer filter in this paper. Firstly, the velocity component of targets is estimated based on the Kalman filter (KF). By using small covariance, the position component of targets is achieved in low-dimensional state space. Besides, the multi-target tracks are systematically merged based on the identifying label. As a result, the running time is saved owing to half of sampling particles. The numeric simulation indicated that the proposed algorithm can track multi-target in clutter circumstance with satisfactory efficiency and reliability.