Multiple Likelihood Models based Particle Filter for Long-term Full Occlusion

Chengjiao Guo, Ying Lu, Takeshi Ikenaga

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Object tracking is one of the most important applications in the field of computer vision. One of the common problems in object tracking is object occlusions. Especially in the presence of long-term full occlusion, or called long-lived full occlusion, during which the target remains invisible for tens of frames, the tracking is more difficult. This paper proposes an occlusion handling scheme based on particle filter. Compared with the conventional particle filter which usually utilizes color as tracking cue, multiple likelihood models: HSV color and gradient orientation likelihoods, are employed in the observation model during occlusion. The incorporation of these two features makes the target distinguishable even if it is occluded by a similar colored object in the background. Also, multiple state noises are introduced to ensure the redetection of the target at the end of full occlusion as well as keeping tracking accuracy under occlusion. Experimental results under different occlusion conditions show that the proposed particle filter achieves robust and accurate performance compared with the particle filter with appearance adaptive models and the color particle filter, even in the condition of long-lived full occlusion.

Original languageEnglish
Pages (from-to)580-589
Number of pages10
JournalJournal of the Institute of Image Electronics Engineers of Japan
Volume39
Issue number5
DOIs
Publication statusPublished - 2010

Fingerprint

Color
Computer vision

Keywords

  • multiple likelihood models
  • multiple state noises
  • object tracking
  • occlusion handling
  • Particle filter

ASJC Scopus subject areas

  • Computer Science (miscellaneous)
  • Electrical and Electronic Engineering

Cite this

Multiple Likelihood Models based Particle Filter for Long-term Full Occlusion. / Guo, Chengjiao; Lu, Ying; Ikenaga, Takeshi.

In: Journal of the Institute of Image Electronics Engineers of Japan, Vol. 39, No. 5, 2010, p. 580-589.

Research output: Contribution to journalArticle

@article{76facdc73f8941898e7a5bc02d0ad56e,
title = "Multiple Likelihood Models based Particle Filter for Long-term Full Occlusion",
abstract = "Object tracking is one of the most important applications in the field of computer vision. One of the common problems in object tracking is object occlusions. Especially in the presence of long-term full occlusion, or called long-lived full occlusion, during which the target remains invisible for tens of frames, the tracking is more difficult. This paper proposes an occlusion handling scheme based on particle filter. Compared with the conventional particle filter which usually utilizes color as tracking cue, multiple likelihood models: HSV color and gradient orientation likelihoods, are employed in the observation model during occlusion. The incorporation of these two features makes the target distinguishable even if it is occluded by a similar colored object in the background. Also, multiple state noises are introduced to ensure the redetection of the target at the end of full occlusion as well as keeping tracking accuracy under occlusion. Experimental results under different occlusion conditions show that the proposed particle filter achieves robust and accurate performance compared with the particle filter with appearance adaptive models and the color particle filter, even in the condition of long-lived full occlusion.",
keywords = "multiple likelihood models, multiple state noises, object tracking, occlusion handling, Particle filter",
author = "Chengjiao Guo and Ying Lu and Takeshi Ikenaga",
year = "2010",
doi = "10.11371/iieej.39.580",
language = "English",
volume = "39",
pages = "580--589",
journal = "Journal of the Institute of Image Electronics Engineers of Japan",
issn = "0285-9831",
publisher = "Institute of Image Electronics Engineers of Japan",
number = "5",

}

TY - JOUR

T1 - Multiple Likelihood Models based Particle Filter for Long-term Full Occlusion

AU - Guo, Chengjiao

AU - Lu, Ying

AU - Ikenaga, Takeshi

PY - 2010

Y1 - 2010

N2 - Object tracking is one of the most important applications in the field of computer vision. One of the common problems in object tracking is object occlusions. Especially in the presence of long-term full occlusion, or called long-lived full occlusion, during which the target remains invisible for tens of frames, the tracking is more difficult. This paper proposes an occlusion handling scheme based on particle filter. Compared with the conventional particle filter which usually utilizes color as tracking cue, multiple likelihood models: HSV color and gradient orientation likelihoods, are employed in the observation model during occlusion. The incorporation of these two features makes the target distinguishable even if it is occluded by a similar colored object in the background. Also, multiple state noises are introduced to ensure the redetection of the target at the end of full occlusion as well as keeping tracking accuracy under occlusion. Experimental results under different occlusion conditions show that the proposed particle filter achieves robust and accurate performance compared with the particle filter with appearance adaptive models and the color particle filter, even in the condition of long-lived full occlusion.

AB - Object tracking is one of the most important applications in the field of computer vision. One of the common problems in object tracking is object occlusions. Especially in the presence of long-term full occlusion, or called long-lived full occlusion, during which the target remains invisible for tens of frames, the tracking is more difficult. This paper proposes an occlusion handling scheme based on particle filter. Compared with the conventional particle filter which usually utilizes color as tracking cue, multiple likelihood models: HSV color and gradient orientation likelihoods, are employed in the observation model during occlusion. The incorporation of these two features makes the target distinguishable even if it is occluded by a similar colored object in the background. Also, multiple state noises are introduced to ensure the redetection of the target at the end of full occlusion as well as keeping tracking accuracy under occlusion. Experimental results under different occlusion conditions show that the proposed particle filter achieves robust and accurate performance compared with the particle filter with appearance adaptive models and the color particle filter, even in the condition of long-lived full occlusion.

KW - multiple likelihood models

KW - multiple state noises

KW - object tracking

KW - occlusion handling

KW - Particle filter

UR - http://www.scopus.com/inward/record.url?scp=84973286482&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84973286482&partnerID=8YFLogxK

U2 - 10.11371/iieej.39.580

DO - 10.11371/iieej.39.580

M3 - Article

AN - SCOPUS:84973286482

VL - 39

SP - 580

EP - 589

JO - Journal of the Institute of Image Electronics Engineers of Japan

JF - Journal of the Institute of Image Electronics Engineers of Japan

SN - 0285-9831

IS - 5

ER -