Reflections on Cognitive Vision Systems

Abstract

A long list of buzzwords which percolated through the computer vision community during the past thirty years leads to the question: does ‘Cognitive Vision Systems’ just denote another such ‘fleeting fad’? Upon closer inspection, many apparent ‘buzzwords’ refer to aspects of computer vision systems which became a legitimate target of widespread research interest due to methodological advances or improvements of computer technology. Following a period during which particular topics had been investigated intensively, associated results merged into the general pool of commonly accepted methods and tools, their preponderance faded and time appeared ‘ripe again for the next buzzword’. Such a non-pejorative use of buzzword in the sense of ‘focus of research attention’ appears appropriate, too, for cognitive vision.

It will be argued that cognitive vision could be characterized by a systematic attempt to conceive and implement computer vision systems based on multiple variably-connected, multi-scale consistency requirements extending beyond the domain of signal and geometric processing into the domain of conceptual representations. This in turn necessitates that methods of formal logic will have to be incorporated into computer vision systems. As a consequence, knowledge has to be explicated in order to facilitate its exploitation in many different contexts.