Object tracking using incremental Fisher discriminant analysis

Ruei Sung Lin, Ming Hsuan Yang, Stephen E. Levinson

Research output: Contribution to journalConference article

Abstract

This paper presents a novel object tracking algorithm using incremental Fisher Linear Discriminant (FLD) algorithm. The sample distribution of the target class is modeled by a single Gaussian and the non-target background class is modeled by a mixture of Gaussians. To a facilitate a multiclass classification problem, we recast the classic FLD algorithm in which the number of classes does not need to be pre-determined. The most discriminant projection matrix that best separates the samples in the projected space is computed using FLD at each frame. Based on the current target location, an efficient sampling algorithm is used to predict the possible locations in the next frame. Using the current projection matrix computed by FLD, the most likely candidate which is closed to the center of the target class in the projected space is selected. Since the FLD is repeatedly computed at each frame, we develop an incremental and efficient method to compute the projection matrix based on the previous results. Experimental results show that our tracker is able to follow the target with large lighting, pose and expression variation.

Original languageEnglish (US)
Pages (from-to)757-760
Number of pages4
JournalProceedings - International Conference on Pattern Recognition
Volume2
StatePublished - Dec 17 2004
EventProceedings of the 17th International Conference on Pattern Recognition, ICPR 2004 - Cambridge, United Kingdom
Duration: Aug 23 2004Aug 26 2004

Fingerprint

Discriminant analysis
Lighting
Sampling

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Object tracking using incremental Fisher discriminant analysis. / Lin, Ruei Sung; Yang, Ming Hsuan; Levinson, Stephen E.

In: Proceedings - International Conference on Pattern Recognition, Vol. 2, 17.12.2004, p. 757-760.

Research output: Contribution to journalConference article

@article{95788fd2addd4315bdd5cb6de2419169,
title = "Object tracking using incremental Fisher discriminant analysis",
abstract = "This paper presents a novel object tracking algorithm using incremental Fisher Linear Discriminant (FLD) algorithm. The sample distribution of the target class is modeled by a single Gaussian and the non-target background class is modeled by a mixture of Gaussians. To a facilitate a multiclass classification problem, we recast the classic FLD algorithm in which the number of classes does not need to be pre-determined. The most discriminant projection matrix that best separates the samples in the projected space is computed using FLD at each frame. Based on the current target location, an efficient sampling algorithm is used to predict the possible locations in the next frame. Using the current projection matrix computed by FLD, the most likely candidate which is closed to the center of the target class in the projected space is selected. Since the FLD is repeatedly computed at each frame, we develop an incremental and efficient method to compute the projection matrix based on the previous results. Experimental results show that our tracker is able to follow the target with large lighting, pose and expression variation.",
author = "Lin, {Ruei Sung} and Yang, {Ming Hsuan} and Levinson, {Stephen E.}",
year = "2004",
month = "12",
day = "17",
language = "English (US)",
volume = "2",
pages = "757--760",
journal = "Proceedings - International Conference on Pattern Recognition",
issn = "1051-4651",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Object tracking using incremental Fisher discriminant analysis

AU - Lin, Ruei Sung

AU - Yang, Ming Hsuan

AU - Levinson, Stephen E.

PY - 2004/12/17

Y1 - 2004/12/17

N2 - This paper presents a novel object tracking algorithm using incremental Fisher Linear Discriminant (FLD) algorithm. The sample distribution of the target class is modeled by a single Gaussian and the non-target background class is modeled by a mixture of Gaussians. To a facilitate a multiclass classification problem, we recast the classic FLD algorithm in which the number of classes does not need to be pre-determined. The most discriminant projection matrix that best separates the samples in the projected space is computed using FLD at each frame. Based on the current target location, an efficient sampling algorithm is used to predict the possible locations in the next frame. Using the current projection matrix computed by FLD, the most likely candidate which is closed to the center of the target class in the projected space is selected. Since the FLD is repeatedly computed at each frame, we develop an incremental and efficient method to compute the projection matrix based on the previous results. Experimental results show that our tracker is able to follow the target with large lighting, pose and expression variation.

AB - This paper presents a novel object tracking algorithm using incremental Fisher Linear Discriminant (FLD) algorithm. The sample distribution of the target class is modeled by a single Gaussian and the non-target background class is modeled by a mixture of Gaussians. To a facilitate a multiclass classification problem, we recast the classic FLD algorithm in which the number of classes does not need to be pre-determined. The most discriminant projection matrix that best separates the samples in the projected space is computed using FLD at each frame. Based on the current target location, an efficient sampling algorithm is used to predict the possible locations in the next frame. Using the current projection matrix computed by FLD, the most likely candidate which is closed to the center of the target class in the projected space is selected. Since the FLD is repeatedly computed at each frame, we develop an incremental and efficient method to compute the projection matrix based on the previous results. Experimental results show that our tracker is able to follow the target with large lighting, pose and expression variation.

UR - http://www.scopus.com/inward/record.url?scp=10044279094&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=10044279094&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:10044279094

VL - 2

SP - 757

EP - 760

JO - Proceedings - International Conference on Pattern Recognition

JF - Proceedings - International Conference on Pattern Recognition

SN - 1051-4651

ER -