Investigating Parallel Interpretation-Tree Model Matching Algorithms with ProSet-Linda

Abstract

This paper discusses the development of algorithms for parallel interpretation-tree model matching for 3-D computer vision applications such as object recognition. The algorithms are developed with a prototyping approach using ProSet-Linda. ProSet is a procedural prototyping language based on the theory of finite sets. The coordination language Linda provides a distributed shared memory model, called tuple space, together with some atomic operations on this shared data space. The combination of both languages, viz. ProSet -Linda, is designed for prototyping parallel algorithms. The classical control algorithm for symbolic data/model matching in computer vision is the Interpretation Tree search algorithm. This algorithm has a high computational complexity when applied to matching problems with large numbers of features. This paper examines parallel variations of this algorithm. Parallel execution can increase the execution performance of model matching, but also make feasible entirely new ways of solving matching problems. In the present paper, we emphasize the development of parallel algorithms with a prototyping approach, not the presentation of performance figures displaying increased performance through parallel execution. The expected improvements attained by the parallel algorithmic variations for interpretation-tree search are analyzed. The implementation of ProSet-Linda is briefly discussed.