Pattern Recognition Using Fractals
D.W.N. Sharp, R. L. While · 1993
We present a novel algorithm for the automatic recognition and classification of fractally-encoded images. The algorithm can recognise images from a fractal transformation library in the presence of scaling, displacement and rotation and is robust in the presence of noise. The algorithm forms the basis for a more general recognition mechanism if non-fractal images are encoded as fractal pictures using Barnsley's Collage Theorem. The algorithm is a member of a recently-discovered novel class of "Communication-Intensive Massively-Parallel" (CIMP) algorithms that have already been used for sorting, tessellation of the plane and pitch-period detection, and which show great promise for use in other application areas.