Dave Horner's Website - Yet another perspective on things...
253 guests
Rough Hits : 4207468
how did u find my site?

A clever person solves a problem. A wise person avoids it. -- Einstein
$$e = \sum_{n=0}^\infty \frac{1}{n!}$$

# Content Based Image Retrieval

Thursday, 12 January 2006 17:09
Content Based Image Retrieval or "CBIR" as it is sometimes called, allows computers to relate one image to another.  CBIR systems can also be used to relate new information to a set of images.  This feature is what is so powerful about CBIR systems.  If you create a CBIR system that allows professionals to annotate it, you can teach the computer to learn what the professionals are seeing within the images themselves.

Right now most CBIR systems are nice because instead of entering text to find images, you provide or click on an existing image to find related ones. This is very useful, but computer scientists are interested in something more.

I dream of a day when the computer can automatically annotate images for me.  When I can create a photo album of all my friends, simply annotate a few images that contain my friend 'john' and from then on out the system automagically clusters all the images that contain 'john'.

Some presentations from the group "LTU tech"
Organising personal pictures with content analysis technology
Finding the Right Image in the Corbis Collection

Fast Multiresolution Image Querying
Multi-feature Hierarchical Template Matching Using Distance Transforms