Regularized Online Mixture of Gaussians for Background Subtraction

Hongbin Wang; Paul Miller

doi:10.1109/AVSS.2011.6027331

Regularized Online Mixture of Gaussians for Background Subtraction

Hongbin Wang
, Paul Miller

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

17 Citations (Scopus)

648 Downloads (Pure)

Abstract

Mixture of Gaussians (MoG) modelling [13] is a popular approach to background subtraction in video sequences. Although the algorithm shows good empirical performance, it lacks theoretical justification. In this paper, we give a justification for it from an online stochastic expectation maximization (EM) viewpoint and extend it to a general framework of regularized online classification EM for MoG with guaranteed convergence. By choosing a special regularization function, l1 norm, we derived a new set of updating equations for l1 regularized online MoG. It is shown empirically that l1 regularized online MoG converge faster than the original online MoG .

Original language	English
Title of host publication	Advanced Video and Signal-Based Surveillance (AVSS), 2011 8th IEEE International Conference on
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	249-254
Number of pages	6
ISBN (Print)	978-1-4577-0845-9
DOIs	https://doi.org/10.1109/AVSS.2011.6027331
Publication status	Published - Sept 2011
Event	8th IEEE International Conference on Advanced Video and Signal-Based Surveillance - Klagenfurt, Austria Duration: 30 Aug 2011 → 02 Sept 2011

Conference

Conference	8th IEEE International Conference on Advanced Video and Signal-Based Surveillance
Country/Territory	Austria
City	Klagenfurt
Period	30/08/2011 → 02/09/2011

ASJC Scopus subject areas

Computer Networks and Communications
Signal Processing

Access to Document

10.1109/AVSS.2011.6027331

Avss2011 final5

Cite this

@inproceedings{142bdc4322c448c5bbb71370a2d54263,

title = "Regularized Online Mixture of Gaussians for Background Subtraction",

abstract = "Mixture of Gaussians (MoG) modelling [13] is a popular approach to background subtraction in video sequences. Although the algorithm shows good empirical performance, it lacks theoretical justification. In this paper, we give a justification for it from an online stochastic expectation maximization (EM) viewpoint and extend it to a general framework of regularized online classification EM for MoG with guaranteed convergence. By choosing a special regularization function, l1 norm, we derived a new set of updating equations for l1 regularized online MoG. It is shown empirically that l1 regularized online MoG converge faster than the original online MoG .",

author = "Hongbin Wang and Paul Miller",

year = "2011",

month = sep,

doi = "10.1109/AVSS.2011.6027331",

language = "English",

isbn = "978-1-4577-0845-9",

pages = "249--254",

booktitle = "Advanced Video and Signal-Based Surveillance (AVSS), 2011 8th IEEE International Conference on",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

address = "United States",

note = "8th IEEE International Conference on Advanced Video and Signal-Based Surveillance ; Conference date: 30-08-2011 Through 02-09-2011",

}

Wang, H & Miller, P 2011, Regularized Online Mixture of Gaussians for Background Subtraction. in Advanced Video and Signal-Based Surveillance (AVSS), 2011 8th IEEE International Conference on. Institute of Electrical and Electronics Engineers Inc., pp. 249-254, 8th IEEE International Conference on Advanced Video and Signal-Based Surveillance, Klagenfurt, Austria, 30/08/2011. https://doi.org/10.1109/AVSS.2011.6027331

TY - GEN

T1 - Regularized Online Mixture of Gaussians for Background Subtraction

AU - Wang, Hongbin

AU - Miller, Paul

PY - 2011/9

Y1 - 2011/9

N2 - Mixture of Gaussians (MoG) modelling [13] is a popular approach to background subtraction in video sequences. Although the algorithm shows good empirical performance, it lacks theoretical justification. In this paper, we give a justification for it from an online stochastic expectation maximization (EM) viewpoint and extend it to a general framework of regularized online classification EM for MoG with guaranteed convergence. By choosing a special regularization function, l1 norm, we derived a new set of updating equations for l1 regularized online MoG. It is shown empirically that l1 regularized online MoG converge faster than the original online MoG .

AB - Mixture of Gaussians (MoG) modelling [13] is a popular approach to background subtraction in video sequences. Although the algorithm shows good empirical performance, it lacks theoretical justification. In this paper, we give a justification for it from an online stochastic expectation maximization (EM) viewpoint and extend it to a general framework of regularized online classification EM for MoG with guaranteed convergence. By choosing a special regularization function, l1 norm, we derived a new set of updating equations for l1 regularized online MoG. It is shown empirically that l1 regularized online MoG converge faster than the original online MoG .

U2 - 10.1109/AVSS.2011.6027331

DO - 10.1109/AVSS.2011.6027331

M3 - Conference contribution

SN - 978-1-4577-0845-9

SP - 249

EP - 254

BT - Advanced Video and Signal-Based Surveillance (AVSS), 2011 8th IEEE International Conference on

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 8th IEEE International Conference on Advanced Video and Signal-Based Surveillance

Y2 - 30 August 2011 through 2 September 2011

ER -

Regularized Online Mixture of Gaussians for Background Subtraction

Abstract

Conference

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this