The polynomial predictive Gaussian mixture MeMBer filter
Jian Jun Yin, Jian Qiu Zhang, Bo Hu, Qi Yong Lu · 2010
We propose a novel multi-target tracking algorithm, called the polynomial predictive Gaussian mixture Multi-target Multi-Bernoulli filter (PPGM-MeMBer) filter. We firstly present a unified state space model where the state equation may describe any dynamics of the true targets, no matter linear or nonlinear and no matter we know them well or not, which is more common in practice. Then we apply the Gaussian mixture MeMBer (GM-MeMBer) filter to the unified model. The analysis results show that the proposed PPGM-MeMBer filter can deal with situations when we do not know the targets dynamics well. The multi-target tracking simulation results verify the effectiveness of the proposed method.