Fast Dual-Graph Regularized Background Foreground Separation

Longxiu Huang; Jing Qin

doi:10.1109/SampTA59647.2023.10301385

Fast Dual-Graph Regularized Background Foreground Separation

Longxiu Huang, Jing Qin

Mathematics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Foreground-background separation is a crucial task in various applications such as computer vision, robotics, and surveillance. Robust Principal Component Analysis (RPCA) is a popular method for this task, which considers the static background as the low-rank component and the moving objects in the foreground as the sparse component. To enhance the performance of RPCA, graph regularization is typically used to incorporate the sophisticated geometry of the background and temporal correlation. However, handling the graph Laplacians can be challenging due to the substantial number of data points. In this study, we propose a novel dual-graph regularized foreground-background separation model based on Sobolev smoothness. Our model is solved using a fast numerical algorithm based on the matrix CUR decomposition. Experimental results on real datasets demonstrate that our proposed algorithm achieves state-of-the-art computational efficiency.

Original language	English
Title of host publication	2023 International Conference on Sampling Theory and Applications, SampTA 2023
ISBN (Electronic)	9798350328851
DOIs	https://doi.org/10.1109/SampTA59647.2023.10301385
State	Published - 2023
Event	2023 International Conference on Sampling Theory and Applications, SampTA 2023 - New Haven, United States Duration: Jul 10 2023 → Jul 14 2023

Publication series

Name	2023 International Conference on Sampling Theory and Applications, SampTA 2023

Conference

Conference	2023 International Conference on Sampling Theory and Applications, SampTA 2023
Country/Territory	United States
City	New Haven
Period	7/10/23 → 7/14/23

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Funding

The research of Qin is supported by the NSF grant DMS-1941197. Huang was partially supported by AMS Simons Travel Grant. This work was supported in part through data provided by Intelligent Robotic Arms Lab at the University of Kentucky, and computational resources and services provided by the Institute for Cyber-Enabled Research at the Michigan State University.

Funders	Funder number
AMS-Simons
Institute for Cyber Enabled Research
National Science Foundation Arctic Social Science Program	DMS-1941197
Michigan State University

Keywords

background foreground separation
CUR decomposition
graph regularization
motion detection
Robust principal component analysis

ASJC Scopus subject areas

Discrete Mathematics and Combinatorics
Statistics and Probability
Artificial Intelligence
Computational Theory and Mathematics
Computer Science Applications
Signal Processing

Access to Document

10.1109/SampTA59647.2023.10301385

Cite this

@inproceedings{3c3f50eb4b514538a1a3b35e18ba4724,

title = "Fast Dual-Graph Regularized Background Foreground Separation",

abstract = "Foreground-background separation is a crucial task in various applications such as computer vision, robotics, and surveillance. Robust Principal Component Analysis (RPCA) is a popular method for this task, which considers the static background as the low-rank component and the moving objects in the foreground as the sparse component. To enhance the performance of RPCA, graph regularization is typically used to incorporate the sophisticated geometry of the background and temporal correlation. However, handling the graph Laplacians can be challenging due to the substantial number of data points. In this study, we propose a novel dual-graph regularized foreground-background separation model based on Sobolev smoothness. Our model is solved using a fast numerical algorithm based on the matrix CUR decomposition. Experimental results on real datasets demonstrate that our proposed algorithm achieves state-of-the-art computational efficiency.",

keywords = "background foreground separation, CUR decomposition, graph regularization, motion detection, Robust principal component analysis",

author = "Longxiu Huang and Jing Qin",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 International Conference on Sampling Theory and Applications, SampTA 2023 ; Conference date: 10-07-2023 Through 14-07-2023",

year = "2023",

doi = "10.1109/SampTA59647.2023.10301385",

language = "English",

series = "2023 International Conference on Sampling Theory and Applications, SampTA 2023",

booktitle = "2023 International Conference on Sampling Theory and Applications, SampTA 2023",

}

Huang, L & Qin, J 2023, Fast Dual-Graph Regularized Background Foreground Separation. in 2023 International Conference on Sampling Theory and Applications, SampTA 2023. 2023 International Conference on Sampling Theory and Applications, SampTA 2023, 2023 International Conference on Sampling Theory and Applications, SampTA 2023, New Haven, United States, 7/10/23. https://doi.org/10.1109/SampTA59647.2023.10301385

TY - GEN

T1 - Fast Dual-Graph Regularized Background Foreground Separation

AU - Huang, Longxiu

AU - Qin, Jing

PY - 2023

Y1 - 2023

N2 - Foreground-background separation is a crucial task in various applications such as computer vision, robotics, and surveillance. Robust Principal Component Analysis (RPCA) is a popular method for this task, which considers the static background as the low-rank component and the moving objects in the foreground as the sparse component. To enhance the performance of RPCA, graph regularization is typically used to incorporate the sophisticated geometry of the background and temporal correlation. However, handling the graph Laplacians can be challenging due to the substantial number of data points. In this study, we propose a novel dual-graph regularized foreground-background separation model based on Sobolev smoothness. Our model is solved using a fast numerical algorithm based on the matrix CUR decomposition. Experimental results on real datasets demonstrate that our proposed algorithm achieves state-of-the-art computational efficiency.

AB - Foreground-background separation is a crucial task in various applications such as computer vision, robotics, and surveillance. Robust Principal Component Analysis (RPCA) is a popular method for this task, which considers the static background as the low-rank component and the moving objects in the foreground as the sparse component. To enhance the performance of RPCA, graph regularization is typically used to incorporate the sophisticated geometry of the background and temporal correlation. However, handling the graph Laplacians can be challenging due to the substantial number of data points. In this study, we propose a novel dual-graph regularized foreground-background separation model based on Sobolev smoothness. Our model is solved using a fast numerical algorithm based on the matrix CUR decomposition. Experimental results on real datasets demonstrate that our proposed algorithm achieves state-of-the-art computational efficiency.

KW - background foreground separation

KW - CUR decomposition

KW - graph regularization

KW - motion detection

KW - Robust principal component analysis

UR - https://www.scopus.com/pages/publications/85178520457

UR - https://www.scopus.com/inward/citedby.url?scp=85178520457&partnerID=8YFLogxK

U2 - 10.1109/SampTA59647.2023.10301385

DO - 10.1109/SampTA59647.2023.10301385

M3 - Conference contribution

AN - SCOPUS:85178520457

T3 - 2023 International Conference on Sampling Theory and Applications, SampTA 2023

BT - 2023 International Conference on Sampling Theory and Applications, SampTA 2023

T2 - 2023 International Conference on Sampling Theory and Applications, SampTA 2023

Y2 - 10 July 2023 through 14 July 2023

ER -

Fast Dual-Graph Regularized Background Foreground Separation

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this