The semantics of an image is a multi-layered concept. At the basic level geometric forms and colours can be extracted from the image and their meaning can be captured in linguistic labels defining objects. The degree to which the identified objects fit the norm may instil another layer of meaning associated with response of human perception to a particular texture, colour, contrast, etc. And finally the relationships that exist among the individual objects in an image, which are defined by the social/cultural/historical context, convey the most abstract message about the image content. At this level of granulation image understanding becomes synonymous with reception of non-visual messages expressed through an image.
One common theme that emerges in this research is the need for the empirical verification of information granulation. This theme in fact unites the knowledge extraction/processing endavours in sciences, engineering and fine arts. Given the profound insights into knowledge processing contributed over the centuries bu artists, philosophers and linguists, it is appropriate to address this investigation as Granular Art. The granular modelling framework will coordinate the following: