
www.Usenet.com
| <-- __Chronological__ --> | <-- __Thread__ --> |
[EMAIL PROTECTED] (Olivier Caignart) wrote in message news:<[EMAIL PROTECTED]>... > Hello ! > > I'm looking for an efficient background subtraction method. I've tried > the Mixture of Gaussians (MOG) and i've found a paper by Kim, Harwood > and Davis, named : > A perturbation method for evaluating background subtraction algorithm. > > In this paper the author test several background segmentation methods > and concluded that the method they have implemented is the best. > Their method is a quantization/clustering method to construct a > background model using kohonen. > > Since i don't have a lot of time to spend implementing this method, i > would like to know where i could find a public source code (C/C++) of > this algorithm. > > Any help, links or suggestion would be welcomed. > > thx > > olivier Hi again, I've read several papers and it appears that i might have a problem even if i succeed in implemtning this algorithm. Their work is based on a pixel codebook based on a "color distorsion metric" with a brigthness ratio. Since i'm working on gray level frames the method cannot work. I would like to know if someone has read this paper, or have an idea on this method. I would also like to know if someone has already heard about background/foreground segmentation using kohonen ? Olivier
| <-- __Chronological__ --> | <-- __Thread__ --> |