Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion

Behzad BOZORGTABAR, Roland GOECKE

Research output: A Conference proceeding or a Chapter in BookConference contribution

Abstract

In this paper, we propose a new visual tracking approach based on the Hierarchical Model Fusion framework, which fuses two different trackers to cope with different tracking problems. We use an Incremental Multiple Principal Component Analysis tracker as our main model as well as an image patch tracker as our auxiliary model. Firstly, we randomly sample image patches within the target region obtained by the main model in the training frames for constructing a visual vocabulary using Histogram of Oriented Gradient features. Secondly, we use a supervised learning algorithm based on a Gaussian Mixture Model, which not only operates on supervised information to improve the discriminative power of the clusters, but also increases the purity of the clusters. Then, auxiliary models are initialised by obtaining confidence scores of image patches based on the similarity between candidates and codewords. In addition, an updating procedure and a result refinement scheme are included in the proposed tracking approach. Experiments on challenging video sequences demonstrate the robustness of the proposed approach to handling occlusion, pose variation and rotation
Original languageEnglish
Title of host publication 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013)
EditorsPaulo de Souza, Ulrich Engelke, Ashfaqur Rahman
Place of PublicationUnited States
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages442-449
Number of pages8
ISBN (Print)9781479921263
DOIs
Publication statusPublished - 2013
Event2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA) - Hobart, Hobart, Australia
Duration: 26 Nov 201328 Nov 2013
http://staff.itee.uq.edu.au/lovell/aprs/dicta13/ (Conference Webpage)

Conference

Conference2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA)
Abbreviated titleDICTA 2013
CountryAustralia
CityHobart
Period26/11/1328/11/13
Internet address

Fingerprint

Fusion reactions
Supervised learning
Electric fuses
Principal component analysis
Learning algorithms
Experiments

Cite this

BOZORGTABAR, B., & GOECKE, R. (2013). Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion. In P. D. Souza, U. Engelke, & A. Rahman (Eds.), 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013) (pp. 442-449). United States: IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/DICTA.2013.6691525
BOZORGTABAR, Behzad ; GOECKE, Roland. / Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion. 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013). editor / Paulo de Souza ; Ulrich Engelke ; Ashfaqur Rahman. United States : IEEE, Institute of Electrical and Electronics Engineers, 2013. pp. 442-449
@inproceedings{47139e358b8542f5bcaa7c21b3ef0c9b,
title = "Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion",
abstract = "In this paper, we propose a new visual tracking approach based on the Hierarchical Model Fusion framework, which fuses two different trackers to cope with different tracking problems. We use an Incremental Multiple Principal Component Analysis tracker as our main model as well as an image patch tracker as our auxiliary model. Firstly, we randomly sample image patches within the target region obtained by the main model in the training frames for constructing a visual vocabulary using Histogram of Oriented Gradient features. Secondly, we use a supervised learning algorithm based on a Gaussian Mixture Model, which not only operates on supervised information to improve the discriminative power of the clusters, but also increases the purity of the clusters. Then, auxiliary models are initialised by obtaining confidence scores of image patches based on the similarity between candidates and codewords. In addition, an updating procedure and a result refinement scheme are included in the proposed tracking approach. Experiments on challenging video sequences demonstrate the robustness of the proposed approach to handling occlusion, pose variation and rotation",
keywords = "Visual Tracking, Incremental Multiple PCA, Hierarchical Model Fusion",
author = "Behzad BOZORGTABAR and Roland GOECKE",
year = "2013",
doi = "10.1109/DICTA.2013.6691525",
language = "English",
isbn = "9781479921263",
pages = "442--449",
editor = "Souza, {Paulo de} and Ulrich Engelke and Ashfaqur Rahman",
booktitle = "2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013)",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
address = "United States",

}

BOZORGTABAR, B & GOECKE, R 2013, Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion. in PD Souza, U Engelke & A Rahman (eds), 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013). IEEE, Institute of Electrical and Electronics Engineers, United States, pp. 442-449, 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA), Hobart, Australia, 26/11/13. https://doi.org/10.1109/DICTA.2013.6691525

Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion. / BOZORGTABAR, Behzad; GOECKE, Roland.

2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013). ed. / Paulo de Souza; Ulrich Engelke; Ashfaqur Rahman. United States : IEEE, Institute of Electrical and Electronics Engineers, 2013. p. 442-449.

Research output: A Conference proceeding or a Chapter in BookConference contribution

TY - GEN

T1 - Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion

AU - BOZORGTABAR, Behzad

AU - GOECKE, Roland

PY - 2013

Y1 - 2013

N2 - In this paper, we propose a new visual tracking approach based on the Hierarchical Model Fusion framework, which fuses two different trackers to cope with different tracking problems. We use an Incremental Multiple Principal Component Analysis tracker as our main model as well as an image patch tracker as our auxiliary model. Firstly, we randomly sample image patches within the target region obtained by the main model in the training frames for constructing a visual vocabulary using Histogram of Oriented Gradient features. Secondly, we use a supervised learning algorithm based on a Gaussian Mixture Model, which not only operates on supervised information to improve the discriminative power of the clusters, but also increases the purity of the clusters. Then, auxiliary models are initialised by obtaining confidence scores of image patches based on the similarity between candidates and codewords. In addition, an updating procedure and a result refinement scheme are included in the proposed tracking approach. Experiments on challenging video sequences demonstrate the robustness of the proposed approach to handling occlusion, pose variation and rotation

AB - In this paper, we propose a new visual tracking approach based on the Hierarchical Model Fusion framework, which fuses two different trackers to cope with different tracking problems. We use an Incremental Multiple Principal Component Analysis tracker as our main model as well as an image patch tracker as our auxiliary model. Firstly, we randomly sample image patches within the target region obtained by the main model in the training frames for constructing a visual vocabulary using Histogram of Oriented Gradient features. Secondly, we use a supervised learning algorithm based on a Gaussian Mixture Model, which not only operates on supervised information to improve the discriminative power of the clusters, but also increases the purity of the clusters. Then, auxiliary models are initialised by obtaining confidence scores of image patches based on the similarity between candidates and codewords. In addition, an updating procedure and a result refinement scheme are included in the proposed tracking approach. Experiments on challenging video sequences demonstrate the robustness of the proposed approach to handling occlusion, pose variation and rotation

KW - Visual Tracking

KW - Incremental Multiple PCA

KW - Hierarchical Model Fusion

U2 - 10.1109/DICTA.2013.6691525

DO - 10.1109/DICTA.2013.6691525

M3 - Conference contribution

SN - 9781479921263

SP - 442

EP - 449

BT - 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013)

A2 - Souza, Paulo de

A2 - Engelke, Ulrich

A2 - Rahman, Ashfaqur

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - United States

ER -

BOZORGTABAR B, GOECKE R. Robust Visual Vocabulary Tracking Using Hierarchical Model Fusion. In Souza PD, Engelke U, Rahman A, editors, 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA 2013). United States: IEEE, Institute of Electrical and Electronics Engineers. 2013. p. 442-449 https://doi.org/10.1109/DICTA.2013.6691525