An integrated framework for adaptive subband image coding

Recent work on filter banks and related expansions has revealed an interesting insight: Different filter bank trees can be regarded as different ways of constructing orthonormal bases for linear signal expansion. In particular, fast algorithms for finding best bases in an operational rate-distortion (R/D) sense have been successfully used in image coding. Independently of this work, recent research has also explored the design of filter banks that optimize energy compaction for a single signal or a class of signals. In this paper, we integrate these two different but complementary approaches to best-basis design and propose a coding paradigm in which subband filters, tree structure, and quantizers are chosen to optimize R/D performance. These coder attributes represent side information. They are selected from a codebook designed off-line from training data, using R/D as the design criterion. This approach provides a rational framework in which to explore alternatives to empirical design of filter banks, quantizers, and other coding parameters. The on-line coding algorithm is a relatively simple extension of current R/Doptimal coding algorithms that operate with fixed filter banks and empirically designed quantizer codebooks. In particular, it is shown that selection of the best adapted filter bank from the codebook is computationally elementary.