NobleBlocks

Research indexThesis

Parallel processing in intermediate-level computer vision

Tremblay

eScholarship@McGill (McGill)1 Jan 1992Open access

NobleIDark:/48914/w1/20261105/8252CCC3DOI10.82308/51012

Abstract

The problem investigated in this thesis is that of giving intermediate-level vision researchers adequate parallel processing tools for their work, where data and computational structures do not fit the SIMD execution model, but require a MIMD execution model instead. The contribution of this thesis is a comparison of 3 general-purpose MIMD parallel processing systems as tools for intermediate-level vision, by evaluating them against criteria which capture the essential issues in programming intermediate-level vision algorithms on such machines. According to my criteria, the best-suited of the 3 systems is composed of the Id functional language running on the Massachusetts Institute of Technology's Tagged-Token Dataflow Architecture (M.I.T. TTDA).

Discussion

More

Suggest a correction or report