Computer vision applications with the associative string processor

Abstract

The ASP (Associative String Processor) is a massively parallel fault-tolerant fully associative processor designed primarily for the implementation of very compact, easily extensible, modular low-MIMD/high-SIMD parallel processing systems capable of supporting real-world applications of continuous data input and tightly integrated numeric and symbolic computations. In this paper, an overview of the ASP architecture is presented, the implementation of a number of computer vision tasks is detailed, and the ASP performance in independent computer vision benchmarks is reported in comparison with other leading parallel processing architectures. The computer vision tasks described illustrate the flexibility of the ASP in supporting the diverse computational requirements (i.e., 2-D arrays, statistics, records, and graphs) of integrated computer vision through the use of the interprocessing element communication network and its global and local associative processing.