Performance of arc consistency algorithms on the CRAY

The consistent labeling problem arises in high level computer vision when assigning semantic meaning to the regions of a n image. One of the drawbacks of this method is that it is rather slow. By using the consistency tests, node, arc and path consistency [9], the search space is drastically reduced. However, for large problems it takes a fair amount of time. To use these algorithms more efficiently, one can take two approaches. First, is to design special purpose hardware to specifically run these algorithms. Second is t o use faster computers. Here again, one can either take advantage of the multiprocessors, which are becoming very widely available, or use supercomputers like the CRAY, CDC, etc. Here, we present results of the performance of these algorithms in the CRAY supercomputer.