Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking

Daniel E. Clark, Tuong V. Ba, Ngu V. Ba

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The problem of jointly detecting whether a target is present in a scene and, if there is, estimating its state can be viewed as a multi-object estimation problem where there is a maximum of one target. This joint detection and estimation problem can be solved using a special case of the multi-object Bayes filter. In this paper we investigate the joint target detection and estimation problem with forward-backward smoothing and propose a sequential Monte Carlo implementation. Finite Set Statistics not only facilitates the development of appropriate joint detection and estimation filters, but also the direct extension of these filtering solutions to their related smoothing counterparts. Preliminary results indicate that using the smoothing has two distinct advantages over just using filtering: Firstly, we are able to more accurately identify the appearance and disappearance of a target in the scene and secondly, we can provide improved state estimates when the target exists. ©2009 ISIF.

Original languageEnglish
Title of host publication2009 12th International Conference on Information Fusion, FUSION 2009
Pages899-906
Number of pages8
Publication statusPublished - 2009
Event2009 12th International Conference on Information Fusion - Seattle, WA, United States
Duration: 6 Jul 20099 Jul 2009

Conference

Conference2009 12th International Conference on Information Fusion
Abbreviated titleFUSION 2009
CountryUnited States
CitySeattle, WA
Period6/07/099/07/09

Fingerprint

Target tracking
Statistics

Keywords

  • Detection
  • Estimation
  • Filtering
  • Smoothing
  • Tracking

Cite this

Clark, D. E., Ba, T. V., & Ba, N. V. (2009). Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking. In 2009 12th International Conference on Information Fusion, FUSION 2009 (pp. 899-906)
Clark, Daniel E. ; Ba, Tuong V. ; Ba, Ngu V. / Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking. 2009 12th International Conference on Information Fusion, FUSION 2009. 2009. pp. 899-906
@inproceedings{8d32c99835a74a04b832e3b08f449ede,
title = "Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking",
abstract = "The problem of jointly detecting whether a target is present in a scene and, if there is, estimating its state can be viewed as a multi-object estimation problem where there is a maximum of one target. This joint detection and estimation problem can be solved using a special case of the multi-object Bayes filter. In this paper we investigate the joint target detection and estimation problem with forward-backward smoothing and propose a sequential Monte Carlo implementation. Finite Set Statistics not only facilitates the development of appropriate joint detection and estimation filters, but also the direct extension of these filtering solutions to their related smoothing counterparts. Preliminary results indicate that using the smoothing has two distinct advantages over just using filtering: Firstly, we are able to more accurately identify the appearance and disappearance of a target in the scene and secondly, we can provide improved state estimates when the target exists. {\circledC}2009 ISIF.",
keywords = "Detection, Estimation, Filtering, Smoothing, Tracking",
author = "Clark, {Daniel E.} and Ba, {Tuong V.} and Ba, {Ngu V.}",
year = "2009",
language = "English",
isbn = "9780982443804",
pages = "899--906",
booktitle = "2009 12th International Conference on Information Fusion, FUSION 2009",

}

Clark, DE, Ba, TV & Ba, NV 2009, Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking. in 2009 12th International Conference on Information Fusion, FUSION 2009. pp. 899-906, 2009 12th International Conference on Information Fusion, Seattle, WA, United States, 6/07/09.

Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking. / Clark, Daniel E.; Ba, Tuong V.; Ba, Ngu V.

2009 12th International Conference on Information Fusion, FUSION 2009. 2009. p. 899-906.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking

AU - Clark, Daniel E.

AU - Ba, Tuong V.

AU - Ba, Ngu V.

PY - 2009

Y1 - 2009

N2 - The problem of jointly detecting whether a target is present in a scene and, if there is, estimating its state can be viewed as a multi-object estimation problem where there is a maximum of one target. This joint detection and estimation problem can be solved using a special case of the multi-object Bayes filter. In this paper we investigate the joint target detection and estimation problem with forward-backward smoothing and propose a sequential Monte Carlo implementation. Finite Set Statistics not only facilitates the development of appropriate joint detection and estimation filters, but also the direct extension of these filtering solutions to their related smoothing counterparts. Preliminary results indicate that using the smoothing has two distinct advantages over just using filtering: Firstly, we are able to more accurately identify the appearance and disappearance of a target in the scene and secondly, we can provide improved state estimates when the target exists. ©2009 ISIF.

AB - The problem of jointly detecting whether a target is present in a scene and, if there is, estimating its state can be viewed as a multi-object estimation problem where there is a maximum of one target. This joint detection and estimation problem can be solved using a special case of the multi-object Bayes filter. In this paper we investigate the joint target detection and estimation problem with forward-backward smoothing and propose a sequential Monte Carlo implementation. Finite Set Statistics not only facilitates the development of appropriate joint detection and estimation filters, but also the direct extension of these filtering solutions to their related smoothing counterparts. Preliminary results indicate that using the smoothing has two distinct advantages over just using filtering: Firstly, we are able to more accurately identify the appearance and disappearance of a target in the scene and secondly, we can provide improved state estimates when the target exists. ©2009 ISIF.

KW - Detection

KW - Estimation

KW - Filtering

KW - Smoothing

KW - Tracking

M3 - Conference contribution

SN - 9780982443804

SP - 899

EP - 906

BT - 2009 12th International Conference on Information Fusion, FUSION 2009

ER -

Clark DE, Ba TV, Ba NV. Forward-backward sequential Monte Carlo smoothing for joint target detection and tracking. In 2009 12th International Conference on Information Fusion, FUSION 2009. 2009. p. 899-906