Resolution scalable image coding with dyadic complementary rational wavelet transforms

Abstract

Traditional wavelet-based image coders use dyadic multiresolution decompositions, resulting in spatial resolutions that are also dyadic in scale, i.e., powers of 1/2 the original resolution. In this paper, we increase the set of decodable spatial resolutions beyond dyadic scales by introducing a two-dimensional wavelet decomposition made of combining wavelet transforms of non-integer dilation factors. We describe how the proposed wavelet decomposition can produce spatial resolutions related to the original resolution by dyadic and non-dyadic ratios, and present how the resulting subband structure can be made compatible with zerotree-based subband coders. © Springer-Verlag Berlin Heidelberg 2010.