i am working on object tracking. i want to use background subtraction. i have tried to find out papers on background subtraction but i am really confused to use which one. I have already done tracking using just gray scale subtraction and thresholding. but now i want to model the background and then use it for tracking. i do not know much about background modelling.So, if anyone can suggest me some techniques and good papers on background subtraction and modelling, i will be greatly thankful.

  • $\begingroup$ Gaussian mixture models for background modelling were once trendy. $\endgroup$
    – user7657
    Jun 8, 2014 at 12:44
  • $\begingroup$ Could you tell us about the papers you have read? I am interested in the subject and would welcome a primer on it. $\endgroup$
    – Jean-Yves
    Jun 8, 2014 at 17:32
  • $\begingroup$ You should provide some more information on the kind of images you're working with. Background subtraction is very specific to your domain. $\endgroup$
    – Phonon
    Jun 9, 2014 at 5:31
  • $\begingroup$ OpenCV has a whole module for background subtraction that you can try with various algorithms. $\endgroup$
    – Adi Shavit
    Jun 9, 2014 at 16:42
  • $\begingroup$ I think that ViBe is a very good and easy to implement background modeling method. But it's patented, so be aware of that if you want to use it. $\endgroup$
    – Micka
    Jun 11, 2014 at 8:55

1 Answer 1


The seminal paper on background modeling is Adaptive background mixture models for real-time tracking by Stauffer and Grimson.

If you have Matlab and the Computer Vision System Toolbox, then you can try using vision.ForegroundDetector, which implements a version of their algorithm.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.