Application of tree structured self-organizing maps in content-based image retrieval

We have developed an image retrieval system, named PicSOM, which uses tree structured self-organizing maps as the method for retrieving images similar to a given set of reference images. A novel technique introduced in the PicSOM system facilitates automatic combination of the responses from multiple TS-SOMs and their hierarchical levels. This mechanism aims at adapting to user's preferences in selecting which images resemble to each other in the particular sense interested by the user. The image queries are performed through the World Wide Web and the queries are iteratively refined as the system exposes more images to the user.