### Background and Foreground Modeling Using an Online EM Algorithm

(2006) The Sixth IEEE International Workshop on Visual Surveillance VS2006. p.9-16- Abstract
- A novel approach to background/foreground segmentation using an online EM algorithm is presented. The method models each layer as a Gaussian mixture, with local, per pixel, parameters for the background layer and global parameters for the foreground layer, utilising information from the entire scene when estimating the foreground. Additionally, the online EM algorithm uses a progressive

learning rate where the relative update speed of each Gaussian component depends on how often the component has been observed. It is shown that the progressive learning rate follows naturally from introduction of a forgetting factor in the log-likelihood.

To reduce the number of mixture components similar foreground components... (More) - A novel approach to background/foreground segmentation using an online EM algorithm is presented. The method models each layer as a Gaussian mixture, with local, per pixel, parameters for the background layer and global parameters for the foreground layer, utilising information from the entire scene when estimating the foreground. Additionally, the online EM algorithm uses a progressive

learning rate where the relative update speed of each Gaussian component depends on how often the component has been observed. It is shown that the progressive learning rate follows naturally from introduction of a forgetting factor in the log-likelihood.

To reduce the number of mixture components similar foreground components are merged using a method based on the Kullback-Leibler distance. A bias is introduced in the variance estimates to avoid the known problem of singularities in the log-likelihood of Gaussian mixtures

when the variance tends to zero.

To allow a decoupling of the learning rate of the Gaussian components and the speed at which stationary objects are incorporated into the background a CUSUM detector is used

instead of the prevailing method that uses the ratio of prior probability to standard deviation.

The algorithm is scale invariant and its properties on gray-scale and RGB videos, as well as on output from an edge detector, is compared to that of another algorithm. Especially for the edge detector video performance increases dramatically. (Less)

Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/709638

- author
- Lindström, Johan
^{LU}; Lindgren, Finn^{LU}; Åström, Karl^{LU}; Holst, Jan^{LU}and Holst, Ulla^{LU} - organization
- publishing date
- 2006
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- adaptive Gaussian mixture, online EM, background subtraction.
- host publication
- IEEE International Workshop on Visual Surveillance
- editor
- Jones, Graeme
- volume
- VS2006
- pages
- 8 pages
- publisher
- Faculty of Computing, Information Systems and mathematics, Kingston University, Surrey, UK
- conference name
- The Sixth IEEE International Workshop on Visual Surveillance
- conference location
- Graz, Austria
- conference dates
- 0001-01-02
- project
- Spatio-Temporal Estimation for Mixture Models and Gaussian Markov Random Fields - Applications to Video Analysis and Environmental Modelling
- language
- English
- LU publication?
- yes
- id
- 58117805-668f-484c-b398-38f0e534e12c (old id 709638)
- date added to LUP
- 2016-04-04 10:54:24
- date last changed
- 2020-12-22 02:16:10

@inproceedings{58117805-668f-484c-b398-38f0e534e12c, abstract = {{A novel approach to background/foreground segmentation using an online EM algorithm is presented. The method models each layer as a Gaussian mixture, with local, per pixel, parameters for the background layer and global parameters for the foreground layer, utilising information from the entire scene when estimating the foreground. Additionally, the online EM algorithm uses a progressive<br/><br> learning rate where the relative update speed of each Gaussian component depends on how often the component has been observed. It is shown that the progressive learning rate follows naturally from introduction of a forgetting factor in the log-likelihood.<br/><br> <br/><br> To reduce the number of mixture components similar foreground components are merged using a method based on the Kullback-Leibler distance. A bias is introduced in the variance estimates to avoid the known problem of singularities in the log-likelihood of Gaussian mixtures<br/><br> when the variance tends to zero.<br/><br> <br/><br> To allow a decoupling of the learning rate of the Gaussian components and the speed at which stationary objects are incorporated into the background a CUSUM detector is used<br/><br> instead of the prevailing method that uses the ratio of prior probability to standard deviation.<br/><br> <br/><br> The algorithm is scale invariant and its properties on gray-scale and RGB videos, as well as on output from an edge detector, is compared to that of another algorithm. Especially for the edge detector video performance increases dramatically.}}, author = {{Lindström, Johan and Lindgren, Finn and Åström, Karl and Holst, Jan and Holst, Ulla}}, booktitle = {{IEEE International Workshop on Visual Surveillance}}, editor = {{Jones, Graeme}}, keywords = {{adaptive Gaussian mixture; online EM; background subtraction.}}, language = {{eng}}, pages = {{9--16}}, publisher = {{Faculty of Computing, Information Systems and mathematics, Kingston University, Surrey, UK}}, title = {{Background and Foreground Modeling Using an Online EM Algorithm}}, volume = {{VS2006}}, year = {{2006}}, }