Modern pattern recognition and the classification of works of art

Abstract

Identification of a visual image by computer may be achieved by ‘training’ the system to recognise examples within determined classification categories. Certain methods of achieving this have been made more feasible (particularly financially) by the advancement of silicon chip technology. This paper explains the underlying principles, with particular reference to the WISARD system developed at Brunei University, and proposes some possible applications to the classification and retrieval of art images.