An adaptive PHD filter for tracking with unknown sensor characteristics
2013 (English)Conference paper (Refereed)
In multi-target tracking, the discrepancy between the nominal and the true values of the model parameters might result in poor performance. In this paper, an adaptive Probability Hypothesis Density (PHD) filter is proposed which accounts for sensor parameter uncertainty. Variational Bayes technique is used for approximate inference which provides analytic expressions for the PHD recursions analogous to the Gaussian mixture implementation of the PHD filter. The proposed method is evaluated in a multi-target tracking scenario. The improvement in the performance is shown in simulations.
Place, publisher, year, edition, pages
2013. 1736-1743 p.
variational Bayes; adaptive filtering; sensor calibration; probability hypothesis density filter; robust filtering; multiple target tracking
IdentifiersURN: urn:nbn:se:liu:diva-100232ISI: 000341370000231ISBN: 978-605-86311-1-3OAI: oai:DiVA.org:liu-100232DiVA: diva2:661053
Information Fusion (FUSION), 2013 16th International Conference on