Particle PHD filter-based multitarget multisensor tracking using FMM and VBEM algorithm

States extraction from the particle probability hypotheses density (PHD) filter is a hotspot in multitarget multisenor tracking research. We find that some clustering algorithms are used to extract the states from the particles. Although the classical finite mixture model (FMM) clustering combined with expectation-maximum (EM) algorithm is better in comparison with other clustering algorithms, it is difficult to deal with the model selection issue. Aimed at this point, we propose a novel multitarget states extraction and target number estimation algorithm for the particle PHD filter. The proposed algorithm makes use of FMM whose parameters and number of components can be derived simultaneously using Variational Bayesian expectation-maximum (VBEM) algorithm. According to the test result, an improvement is made in the states extraction based on FMM and VBEM algorithm. Experimental simulations show that the proposed algorithm is more effective in the extraction of states and the estimation of target number.