ABSTRACT
This chapter discusses the use of blackboard architecture in capturing both the general knowledge about the visual world and, at the same time, the high-level knowledge available for tackling the specialized domain at hand. The blackboard architecture is composed of three parts: a blackboard data structure, a set of knowledge sources, and a control mechanism. The features of the blackboard architecture with the extensions of blackboard dimensions, frame-based object models, and parallel activation of knowledge sources provide the necessary framework for integrating these knowledge sources. A computer system for the automatic interpretation of a television camera signal relies on knowledge of the world. The development of powerful mathematical models of the process of image formation and its relation to surface shape and other object characteristics is an important area for research in computer vision. Research in restricted domains, such as medical and industrial image interpretation, will lead to useful applications.
