Self-similarity clustering of random texture via stochastic-computational complexity analysis

A nondeterministic scheme is presented for self-similar clustering of random texture. By modeling observed texture as the attractor associated with unknown contraction mappings, a capturing probability is induced on the image plane. Guided by maximum entropy growth of discrete stochastic features, the statistics of the mapping range is evaluated. Variance analysis is applied to estimate mapping parameters for partitioning the texture pattern into subregions of a fractal attractor. The proposed scheme was implemented and verified through simulation studies