Prototyping a parallel vision system in Standard ML

Abstract

The construction of a parallel vision system from Standard ML prototypes is presented. The system recognises 3D objects from 2D scenes through edge detection, grouping of edges into straight lines and line junction based model matching. Functional prototyping for parallelism is illustrated through the development of the straight line detection component. The assemblage of the whole system from prototyped components is then considered and its performance discussed.

A functional methodology for parallel image processing development

“…In short, the application should be portable across a variety of parallel architectures and indeed generations of the same parallel architecture where conventionally a upgrade would mean a new tuning phase. This moves forward from existing approaches, such as that of Michaelson and Scaife [3] who prototyped image processing systems in sequential ML but ported to and parallelised with occam. Here the parallelisation forms such an integral part of the prototyping process, that it can be abstracted in a flexible functional manner and re-used composably.…”

Prototyping Parallel Algorithms using Standard ML.

“…From previous work [9,3] we have identified a small set of useful HOFs for which we have efficient parallel implementations. These are:…”

“…We give brief examples of each algorithm in SML using the higher order function constructs that we have indicated can be implemented in parallel. Although we do not explicitly state, for each example, how the parallel implementation is derived from the prototype, we have substantial experience in this process [9,11] using 0ccam2 as the target language based on a Meiko transputer surface. We have shown that if we can write a functional language program using our standard set of higher order functions then we can develop an equivalent parallel implementation based on the parallel constructs for these functions.…”

“…We have studied the Hough transform and have implemented a straight line detector in SML and 0ccam2 [9] …”

Automatic verification of functions with accumulating parameters

“…We are partly motivated by a research project 1 in which a parallelizing compiler for Standard ML (SML) is being developed. This project builds directly upon previous work on the development of parallel systems from functional prototypes (Michaelson and Scaife, 1995). Transformation rules for SML will play an important part within the compilation process.…”