Recognition of fractal images using a neural network

Abstract

In this paper we present a neural network that is able to determine whether a given pixel image represents the visualization of a fractal structure or not. The proposed network is a hierarchically organized, multi-level feedforward architecture which has been designed to exploit the structural properties of artificially generated fractals. The basic idea is to extract the generator of a fractal image and train the network via backpropagation to produce the correct classification. The classification quality of the network is tested on several images, both fractal with/without noise and non-fractal, and it will be demonstrated that the network is able to correctly classify the test images up to a certain signal-to-noise ratio. An efficient parallel implementation of the network on a multi-transputer system is described.