A REVIEW OF VARIOUS PHILOSOPHIES ON THE DESIGN AND ANALYSIS OF PARALLEL COMPUTER ARCHITECTURES FOR VISION

Parallel computer vision is a key element in many applications, where intelligent devices are needed Several classes of synthesised and proposed vision-oriented parallel computer architectures are reviewed. Key design principles and methodologies of these representative and state-of-the-art systems are used to identify and classify various properties of such systems. We will show how a study of the philosophies used in the designs of various systems can result in a classification of such parallel computers. These results are then used to outline and analyse some specific challenges and trade-offs that must be considered in the design and analysis of vision-oriented parallel computing.