A model for image bases and its query facility

A model for image bases is presented, along with a query facility for exploiting image bases contents. The models views an image as a 3-level object, comprised of a form, a content and a mapping component. The form level provides an objective model for images as syntactical objects, against which visual queries (Le. queries being images themselves) can be stated. The content level provides a representation of the scene depicted by an image, in a semantic data modelling style. The mapping level serves to connect the form and content levels, establishing a relationship between visual appearances of individuals and their conceptual descriptions. A query is a sentence of a logical language that can address any subset of the three image components and also the spatial relationships between objects of the form level (i.e. image regions) or of the content level (Le. content entities). An image is retrieved in response to a query if it logically satisfies the query.