Continuous-tone image recognition using fractal theory
Ying Liu · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1993
In this paper, we study pattern recognition using Probabilistic Iterated Function Systems (PIFS). A learning system can be defined by three rules: the encoding rule, the rule of internal change, and the quantization rule. In our system, the data encoding is to store an image in a stable distribution of a PIFS. Given an input image f (epsilon) F, one can find a PIFS t (epsilon) T such that the equilibrium distribution of this PIFS is the given image f. Therefore, the input image, f, is encoded into a specification of a PIFS, t. This mapping from F (image space) to T (parameter space of PIFS) defines fractal transformation. Fractal transformation encodes an input image into a relatively small vector which catches the characteristics of the input vector. The internal space T is the parameter space of PIFS. The internal change rule of our system uses a local minima algorithm to encode the input data. The output data of the encoding stage is a specification of a stochastic dynamical system. The quantization rule divides the internal data space T by sample data.