Distributed computing for vision: Architecture and preliminary results.

One way of speeding up algorithms is to use distributed vision the use of more than one processor. This paper first describes a system at the latter end: a system of multiprocessors on a high-speed bus. It then describes a computer vision algorithm for object location that lends itself to distribution on such a system. Performance data for this algorithm was collected from implementations on a distributed network of microcomputers, a Vax 11/750, and a Vax 11/780. These results indicate the usefulness of this approach. Finally, other more interesting distributed algorithms are described and discussed.