Distributed Computing for Vision: Architecture and a Benchmark Test

Abstract

Computer vision algorithms are notorious for their computational expense. Distributed vision, the use of more than one processor, can decrease computation costs and speed up algorithms. There are various ways to do this, ranging from parallelism at the sensor level to true multiprocessor systems. This correspondence first describes a system of the latter type: a system of microprocessors on a high-speed bus. A canonical vision task, locating a number of objects and measuring certain two-dimensional features of those objects, serves as a benchmark test for the system. An algorithm for this task is presented. Performance measures are compared from implementations on the distributed system, a Vax 11/750, and a Vax 11/780. Results indicate that three microprocessors outperform a Vax 11/780 at this task. Finally, other more interesting distributed algorithms are briefly discussed.