DocumentCode :
3656870
Title :
Distributed multi-target tracking via generalized multi-Bernoulli random finite sets
Author :
Bailu Wang; Wei Yi; Suqi Li;Mark R. Morelande; Lingjiang Kong; Xiaobo Yang
Author_Institution :
Sch. of Electron. Eng., Univ. of Electron. Sci. &
fYear :
2015
fDate :
7/1/2015 12:00:00 AM
Firstpage :
253
Lastpage :
261
Abstract :
In this paper, we address the problem of the distributed multi-target tracking with labelled set filters in the framework of generalized Covariance Intersection (GCI). Our analyses show that the label space mismatching phenomenon, which means the same realization drawn from label spaces of different sensors does not have the same implication, is quite common in practical scenarios and may bring serious problems. To get rid of the bad influence of label space mismatching phenomenon, firstly, we propose a robust strategy for distributed fusion with labelled set posteriors in which labelled set posteriors are transformed to their unlabelled versions firstly and the GCI fusion is performed with the unlabelled posteriors then. Secondly, we derive the unlabelled versions of common labelled set distributions in generalized labelled multi-Bernoulli (GLMB) family and show that they all belong to the same (unlabelled) random finite set (RFS) family, referred to as generalized multi-Bernoulli (GMB) family. Thirdly, we derive the explicit formula for GCI with GMB distributions, which enables the distributed fusion with GLMB filter family, including the GLMB, δ-GLMB, Mδ-GLMB and LMB filters. Simulation results for Gaussian mixture implementation have demonstrated the performance of the proposed distributed fusion algorithms in two challenging tracking scenarios.
Keywords :
"Sensor phenomena and characterization","Sensor fusion","Extraterrestrial phenomena","Robustness","Approximation methods","Bayes methods"
Publisher :
ieee
Conference_Titel :
Information Fusion (Fusion), 2015 18th International Conference on
Type :
conf
Filename :
7266570
Link To Document :
بازگشت