Improved fractal-based methods for image compression

Abstract

On the basis of the theory of iterated function systems (IFS), an image may be represented by a dynamic system, known as a fractal transformation, whose attractor is close to the image to be coded. For natural images, the parameters of such a dynamic system can usually be coded very compactly, which makes this method suitable for image compression. In the fractal based compression methods, the problem of coding an image is to design such a dynamic system. To do this, an image is first partitioned into segments and each is approximated, in part, by another larger segment within the same image. These coding methods make use of the redundancies in the image at different scales. The decoding is then done by finding the attractor of the dynamic system and is typically an iterated process.