[PDF] Heterogeneous Unlabeled and Labeled RFS Filter Fusion for Scalable Multisensor Multitarget Tracking-论文阅读讨论-ReadPaper

摘要

This paper proposes a heterogenous density fusion approach to scalablemultisensor multitarget tracking where the local, inter-connected sensors rundifferent types of random finite set (RFS) filters according to theirrespective capacity and need. They result in heterogenous multitarget densitiesthat are to be fused with each other in a proper means for more robust andaccurate detection and localization of the targets. Our recent work has exposeda key common property of effective arithmetic average (AA) fusion approaches toboth unlabeled and labeled RFS filters which are all built on averaging theirrelevant un-labeled/labeled probability hypothesis densities (PHDs). Thanks tothis, this paper proposes the first ever heterogenous unlabeled and labeled RFSfilter cooperation approach based on Gaussian mixture implementations where thelocal Gaussian components (L-GCs) are so optimized that the resulting unlabeledPHDs best fit their AA, regardless of the specific type of the local densities.To this end, a computationally efficient, approximate approach is proposedwhich only revises the weights of the L-GCs, keeping the other parameters ofL-GCs unchanged. In particular, the PHD filter, the unlabeled and labeledmulti-Bernoulli (MB/LMB) filters are considered. Simulations have demonstratedthe effectiveness of the proposed approach for both homogeneous andheterogenous fusion of the PHD-MB- LMB filters in different configurations.

展开全部

图表提取

论文十问

参考文献

被引用

提问与回答

暂无人提供速读十问回答