Generalized fractal-wavelet transforms for image representation and compression

Abstract

In this paper a set of generalized 2D fractal-wavelet transforms is introduced. Their primary difference from the usual transforms lies in treating "horizontal", "vertical" and "diagonal" quadtrees independently. Although more parameters are required, the added flexibility makes it quite tractable for image compression. Even the possibility of independent scaling factors but common parent blocks for the three sets of quadtrees gives excellent approximations with quite high compression ratios.