PHD filter
Attached is the M code for the SMC-PHD filter for a near constant velocity multi-target tracking model, where the multi-estimate extraction (MEE) is carried out by using the so-called MEAP (multi-EAP) method, see
T. Li, J. M. Corchado, S. Sun and H. Fan, Multi-EAP: extended EAP for multi-estimate extraction for the SMC-PHD filter, Chinese Journal of Aeronautics, vol.30, no.1, pp. 368–379, 2017. OPEN ACCESS
Random finite set-based multi-target density estimator graciously avoids the observation-to-track association difficulty in the filtering process, but leaves no clues of how to optimally extract state-estimates of multiple targets from the multi-target density, namely multi-estimate extraction (MEE). MEE is an essential requirement for the multi-target tracker, where its key performance assessment is based on the accuracy, computational efficiency and reliability of the solution. However, it is particularly challenging for the sequential Monte Carlo implementation of the probability hypothesis density filter.
Here, in my implementation, the MEE problem is formulated approximately as a family of parallel single-estimate extraction problems, where the optimal Expected a Posteriori (EAP) estimator is employed. The present multi-Expected a Posteriori (MEAP) estimator is free of iterative clustering computation and yields accurate and reliable estimation that is approximately optimal in the sense that each EAP estimator minimizes the mean square error of its estimation. Typical simulation models are employed to demonstrate the superiority of the MEAP estimator over the state-of-the-art methods in terms of fast processing speed and high estimation accuracy.
More importantly, the MEE is very suitable for parallel processing. Based on it, a novel parallel processing framework of the SMC-PHD filter is developed, which is significantly different to the traditional parallel processing of the SMC/PF filters.....See:
T. Li, S. Sun, M. Bolic and J. M. Corchado. Algorithm design for parallel implementation of the SMC-PHD filter, Signal Processing, 2016, vol.119, pp. 115-127. @ ScienceDirect