Fractal encoding of images by classified domain trees

Fractal compression of digital images has attracted much attention. It is based on the mathematical theory of iterated function systems (IFS) developed by Hutchinson (1981) and Barnsley (1988). We show that the tree method may be combined with the block classification method to get a greater speed-up. In block classification the classes are stored without any additional structure, usually as a list, and searched linearly until a best match is found. We should expect an enhancement if we arrange the classes into trees, and use the pruning algorithm for the search. Another way to express this idea is that the domain pool is grown into lots of small trees rather than one big one. The experiments show that a significant speed-up is achievable provided two conditions are met. First, only reasonable size classes should be grown into a tree. Second, the feature extraction methods for the classification and for the tree construction should be distinct