Generalized formulation and hypercube algorithms for relaxationlabeling

摘要

Presents a generalized formulation for several well-knownapproaches to relaxation labeling, including discrete, fuzzy, linearprobabilistic models and several nonlinear probabilistic modes. Based onthis generalized framework, two parallel algorithms for SIMD hypercubecomputers with different numbers of processors are proposed andanalyzed. The algorithms achieve minimal time complexity