An analysis of a distributed multiresolution vision system

Abstract This paper analyzes a distributed vision system that uses a multiresolution imagery to achieve computational savings in dynamic scene analysis. The system basically contains a set of processes that asynchronously work on a pyramid image structure. The system also contains a blackboard that provides inter-process communications as well as an overall scene interpretation. The paper examines the capability of each component in the system and proposes an underlying architecture. Finally, a performance evaluation is provided by comparing the present system with a conventional vision system.