Re: [help-GIFT] Re: Clarification on inverted file

[suryani lim]

Hi Wolfgang,

Thanks for the information on GIFT and the references.
Thanks also go to David for your comments.

> If I may add to that, Smeulders et al. report in their recent
> overview article some work done in 1996 that was about 
> indexing images in inverted files using salient points. 

Are you refering to this article?
Content-Based Image Retrieval at the End of the Early Years 
by  Arnold W.M. Smeulders, Marcel Worring, Simone Santini,
Amarnath Gupta, Ramesh Jain
appeared in IEEE transaction on Pattern Analysis and Machine
Intelligence December 2000 issues?

> Chad Carson told me, that he stored the blobs of
> BlobWorld in an inverted file. 

I did a search and found this article:

Blobworld: A system for region-based image indexing and retrieval 
at
http://www.cs.berkeley.edu/~carson/papers/visual99.html

I suppose Chad Carson was referring to this work.

> MARS is strongly inspired by 
> text retrieval, but modifies the retrieval scheme, basing the
> weighting not on the document frequency but on the standard
> deviation of the term frequency.

I had a look at MARS home page at
http://www-db.ics.uci.edu/pages/research/mars.shtml

but I don't know how many MARS there are. Are you talking about that
MARS?

> After what I have seen, David's approach that was implemented 
> in Viper is maybe the most radical approach to adapting image
> retrieval methods to text retrieval indexing, because it does not
> try to achieve semantically meaningful representation (e.g.
> segments) before indexing the image in an
> inverted file. So much so, that this inverted files are not 

Right. 

> More questions welcome...

No regrets? ;-)

Suryani.