High-order context modeling and embedded conditional entropy coding of wavelet coefficients for image compression

Much of recent progress in wavelet image compression came from better context modeling and entropy coding of quantized wavelet coefficients. In the past few months many papers on the subject were published. They reported rate-distortion performance results that are the best or near the best in the literature. Given seemingly ever smaller improvements on R-D performance with increasing sophistication and complexity of current R-D optimized quantizers and context models of wavelet coefficients, two tantalizing questions are whether there still exists some room for even higher coding efficiency of wavelet image coders, and if so, whether the improvement can be made with an embedded code stream. This paper will shed some lights on, and offer modestly encouraging answers to these questions.