Efficient implementation of multilevel algorithms on hypercube supercomputers for computer vision

摘要: The authors deal with the problem of implementing multilevel algorithms on hypercube parallel computers for computer vision problems. conventional pyramid is a structure identical 2 /spl times/ reductions between all pairs neighboring levels. mapping proposed source structures and target topology are based some most important algorithms. Application classified according to their performance requirements appropriate then chosen implementation new compared set graph embedding metrics from results produced Connection Machine system CM-2 massively computer. comparison shows that selection other than more efficient solution time. In addition, algorithm often becomes very critical decision high yields.