Parallel algorithms for computer vision. Contract report, 31 August 1985-31 August 1986

The general goals of this research effort is to explore the potential applications and performance of fine-grained computer architectures for vision. The body of this report gives a brief overview of the results of the research during the first twelve months of effort. A taxonomy of parallel vision algorithms was refined. Parallel versions of several Class 1 algorithms (for early vision, directly derived from regularization methods) were designed. This preparatory work allowed the implementation of several of these algorithms on the Connection Machine within a few days when it became available to the researchers. Class 2 and 3 algorithms have also been designed, while at the same time studying some of the associated basic problems in recognition and representation.