Object tracking via partial least squares analysis

Qing Wang, Feng Chen, Wenli Xu, Ming Hsuan Yang

Research output: Contribution to journalArticle

113 Citations (Scopus)

Abstract

We propose an object tracking algorithm that learns a set of appearance models for adaptive discriminative object representation. In this paper, object tracking is posed as a binary classification problem in which the correlation of object appearance and class labels from foreground and background is modeled by partial least squares (PLS) analysis, for generating a low-dimensional discriminative feature subspace. As object appearance is temporally correlated and likely to repeat over time, we learn and adapt multiple appearance models with PLS analysis for robust tracking. The proposed algorithm exploits both the ground truth appearance information of the target labeled in the first frame and the image observations obtained online, thereby alleviating the tracking drift problem caused by model update. Experiments on numerous challenging sequences and comparisons to state-of-the-art methods demonstrate favorable performance of the proposed tracking algorithm.

Original languageEnglish
Article number6224182
Pages (from-to)4454-4465
Number of pages12
JournalIEEE Transactions on Image Processing
Volume21
Issue number10
DOIs
Publication statusPublished - 2012 Sep 28

Fingerprint

Labels
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Graphics and Computer-Aided Design

Cite this

Wang, Qing ; Chen, Feng ; Xu, Wenli ; Yang, Ming Hsuan. / Object tracking via partial least squares analysis. In: IEEE Transactions on Image Processing. 2012 ; Vol. 21, No. 10. pp. 4454-4465.
@article{7a669494b6e24d8a89f9f1b456719102,
title = "Object tracking via partial least squares analysis",
abstract = "We propose an object tracking algorithm that learns a set of appearance models for adaptive discriminative object representation. In this paper, object tracking is posed as a binary classification problem in which the correlation of object appearance and class labels from foreground and background is modeled by partial least squares (PLS) analysis, for generating a low-dimensional discriminative feature subspace. As object appearance is temporally correlated and likely to repeat over time, we learn and adapt multiple appearance models with PLS analysis for robust tracking. The proposed algorithm exploits both the ground truth appearance information of the target labeled in the first frame and the image observations obtained online, thereby alleviating the tracking drift problem caused by model update. Experiments on numerous challenging sequences and comparisons to state-of-the-art methods demonstrate favorable performance of the proposed tracking algorithm.",
author = "Qing Wang and Feng Chen and Wenli Xu and Yang, {Ming Hsuan}",
year = "2012",
month = "9",
day = "28",
doi = "10.1109/TIP.2012.2205700",
language = "English",
volume = "21",
pages = "4454--4465",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "10",

}

Object tracking via partial least squares analysis. / Wang, Qing; Chen, Feng; Xu, Wenli; Yang, Ming Hsuan.

In: IEEE Transactions on Image Processing, Vol. 21, No. 10, 6224182, 28.09.2012, p. 4454-4465.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Object tracking via partial least squares analysis

AU - Wang, Qing

AU - Chen, Feng

AU - Xu, Wenli

AU - Yang, Ming Hsuan

PY - 2012/9/28

Y1 - 2012/9/28

N2 - We propose an object tracking algorithm that learns a set of appearance models for adaptive discriminative object representation. In this paper, object tracking is posed as a binary classification problem in which the correlation of object appearance and class labels from foreground and background is modeled by partial least squares (PLS) analysis, for generating a low-dimensional discriminative feature subspace. As object appearance is temporally correlated and likely to repeat over time, we learn and adapt multiple appearance models with PLS analysis for robust tracking. The proposed algorithm exploits both the ground truth appearance information of the target labeled in the first frame and the image observations obtained online, thereby alleviating the tracking drift problem caused by model update. Experiments on numerous challenging sequences and comparisons to state-of-the-art methods demonstrate favorable performance of the proposed tracking algorithm.

AB - We propose an object tracking algorithm that learns a set of appearance models for adaptive discriminative object representation. In this paper, object tracking is posed as a binary classification problem in which the correlation of object appearance and class labels from foreground and background is modeled by partial least squares (PLS) analysis, for generating a low-dimensional discriminative feature subspace. As object appearance is temporally correlated and likely to repeat over time, we learn and adapt multiple appearance models with PLS analysis for robust tracking. The proposed algorithm exploits both the ground truth appearance information of the target labeled in the first frame and the image observations obtained online, thereby alleviating the tracking drift problem caused by model update. Experiments on numerous challenging sequences and comparisons to state-of-the-art methods demonstrate favorable performance of the proposed tracking algorithm.

UR - http://www.scopus.com/inward/record.url?scp=84866604392&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84866604392&partnerID=8YFLogxK

U2 - 10.1109/TIP.2012.2205700

DO - 10.1109/TIP.2012.2205700

M3 - Article

AN - SCOPUS:84866604392

VL - 21

SP - 4454

EP - 4465

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

IS - 10

M1 - 6224182

ER -