The terminological image retrieval model

Abstract

We present a model for image retrieval in which images are represented both at the form level, as sets of physical features of the representing objects, and at the content level, as sets of logical assertions about the represented entities as well as about facts of the subject matter that are deemed as relevant for retrieval. A uniform and powerful query language allows queries to be issued that transparently combine features pertaining to form and content. Queries are expressions of a fuzzy logical language. While that part of the query that pertains to (medium-independent) content is "directly" processed by an inferential engine, that part that pertains to (medium-dependent) form is entrusted to specialised signal processing procedures linked to the logical language by a procedural attachment mechanism.