Dynamic texture as foreground and background

D. Chetverikov, Sándor Fazekas, Michal Haindl

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

Depending on application, temporal texture can be viewed as either foreground or background. We address two related problems: finding regions of dynamic texture in a video and detecting moving targets in a dynamic texture. We propose efficient and fast methods for both cases. The methods can be potentially used in real-time applications of machine vision. First, we show how the optical flow residual can be used to find dynamic texture in video. The algorithm is a practical, real-time simplification of the sophisticated and powerful but time-consuming method (Fazekas et al. in Int J Comput Vis 82:48-63, 2009). We give numerous examples of detecting and segmenting fire, smoke, water and other dynamic textures in real-world videos acquired by static and moving cameras. Then we apply the singular value decomposition (SVD) to a temporal datawindowin a video to detect targets in dynamic texture via the residual of the largest singular value. For a dynamic background of low-temporal periodicity, such as water, no temporal periodicity analysis is needed. For a highly periodic background such as an escalator, we show that periodicity analysis can improve detection results. Applying the method proposed in Chetverikov and Fazekas (Proceedings of British machine vision conference, vol 1, pp 167-176, 2006), we find the temporal period and use the resonant SVD to detect moving targets against a timeperiodic background.

Original languageEnglish
Pages (from-to)741-750
Number of pages10
JournalMachine Vision and Applications
Volume22
Issue number5
DOIs
Publication statusPublished - Sep 2011

Fingerprint

Textures
Singular value decomposition
Computer vision
Escalators
Optical flows
Smoke
Water
Fires
Cameras

Keywords

  • Background modelling
  • Detection
  • Dynamic texture
  • Optical flow
  • Photometric invariants
  • SVD
  • Temporal periodicity

ASJC Scopus subject areas

  • Hardware and Architecture
  • Computer Vision and Pattern Recognition
  • Software
  • Computer Science Applications

Cite this

Dynamic texture as foreground and background. / Chetverikov, D.; Fazekas, Sándor; Haindl, Michal.

In: Machine Vision and Applications, Vol. 22, No. 5, 09.2011, p. 741-750.

Research output: Contribution to journalArticle

Chetverikov, D. ; Fazekas, Sándor ; Haindl, Michal. / Dynamic texture as foreground and background. In: Machine Vision and Applications. 2011 ; Vol. 22, No. 5. pp. 741-750.
@article{9339ffab67fe4259b9a382b116a1cd20,
title = "Dynamic texture as foreground and background",
abstract = "Depending on application, temporal texture can be viewed as either foreground or background. We address two related problems: finding regions of dynamic texture in a video and detecting moving targets in a dynamic texture. We propose efficient and fast methods for both cases. The methods can be potentially used in real-time applications of machine vision. First, we show how the optical flow residual can be used to find dynamic texture in video. The algorithm is a practical, real-time simplification of the sophisticated and powerful but time-consuming method (Fazekas et al. in Int J Comput Vis 82:48-63, 2009). We give numerous examples of detecting and segmenting fire, smoke, water and other dynamic textures in real-world videos acquired by static and moving cameras. Then we apply the singular value decomposition (SVD) to a temporal datawindowin a video to detect targets in dynamic texture via the residual of the largest singular value. For a dynamic background of low-temporal periodicity, such as water, no temporal periodicity analysis is needed. For a highly periodic background such as an escalator, we show that periodicity analysis can improve detection results. Applying the method proposed in Chetverikov and Fazekas (Proceedings of British machine vision conference, vol 1, pp 167-176, 2006), we find the temporal period and use the resonant SVD to detect moving targets against a timeperiodic background.",
keywords = "Background modelling, Detection, Dynamic texture, Optical flow, Photometric invariants, SVD, Temporal periodicity",
author = "D. Chetverikov and S{\'a}ndor Fazekas and Michal Haindl",
year = "2011",
month = "9",
doi = "10.1007/s00138-010-0251-6",
language = "English",
volume = "22",
pages = "741--750",
journal = "Machine Vision and Applications",
issn = "0932-8092",
publisher = "Springer Verlag",
number = "5",

}

TY - JOUR

T1 - Dynamic texture as foreground and background

AU - Chetverikov, D.

AU - Fazekas, Sándor

AU - Haindl, Michal

PY - 2011/9

Y1 - 2011/9

N2 - Depending on application, temporal texture can be viewed as either foreground or background. We address two related problems: finding regions of dynamic texture in a video and detecting moving targets in a dynamic texture. We propose efficient and fast methods for both cases. The methods can be potentially used in real-time applications of machine vision. First, we show how the optical flow residual can be used to find dynamic texture in video. The algorithm is a practical, real-time simplification of the sophisticated and powerful but time-consuming method (Fazekas et al. in Int J Comput Vis 82:48-63, 2009). We give numerous examples of detecting and segmenting fire, smoke, water and other dynamic textures in real-world videos acquired by static and moving cameras. Then we apply the singular value decomposition (SVD) to a temporal datawindowin a video to detect targets in dynamic texture via the residual of the largest singular value. For a dynamic background of low-temporal periodicity, such as water, no temporal periodicity analysis is needed. For a highly periodic background such as an escalator, we show that periodicity analysis can improve detection results. Applying the method proposed in Chetverikov and Fazekas (Proceedings of British machine vision conference, vol 1, pp 167-176, 2006), we find the temporal period and use the resonant SVD to detect moving targets against a timeperiodic background.

AB - Depending on application, temporal texture can be viewed as either foreground or background. We address two related problems: finding regions of dynamic texture in a video and detecting moving targets in a dynamic texture. We propose efficient and fast methods for both cases. The methods can be potentially used in real-time applications of machine vision. First, we show how the optical flow residual can be used to find dynamic texture in video. The algorithm is a practical, real-time simplification of the sophisticated and powerful but time-consuming method (Fazekas et al. in Int J Comput Vis 82:48-63, 2009). We give numerous examples of detecting and segmenting fire, smoke, water and other dynamic textures in real-world videos acquired by static and moving cameras. Then we apply the singular value decomposition (SVD) to a temporal datawindowin a video to detect targets in dynamic texture via the residual of the largest singular value. For a dynamic background of low-temporal periodicity, such as water, no temporal periodicity analysis is needed. For a highly periodic background such as an escalator, we show that periodicity analysis can improve detection results. Applying the method proposed in Chetverikov and Fazekas (Proceedings of British machine vision conference, vol 1, pp 167-176, 2006), we find the temporal period and use the resonant SVD to detect moving targets against a timeperiodic background.

KW - Background modelling

KW - Detection

KW - Dynamic texture

KW - Optical flow

KW - Photometric invariants

KW - SVD

KW - Temporal periodicity

UR - http://www.scopus.com/inward/record.url?scp=80053895602&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80053895602&partnerID=8YFLogxK

U2 - 10.1007/s00138-010-0251-6

DO - 10.1007/s00138-010-0251-6

M3 - Article

VL - 22

SP - 741

EP - 750

JO - Machine Vision and Applications

JF - Machine Vision and Applications

SN - 0932-8092

IS - 5

ER -