Abstract

In this work, we propose end-to-end image compression /decompression systems based on the discrete wavelet transform (DWT), entropy-constrained trellis coded quantization (ECTCQ), and arithmetic coding. Two forms of adaptation (codebook and image subblock) are investigated. The baseline system uses neither form of adaptation, while the adaptive systems use one or both forms of adaptivity. Excellent peak signal-to-noise ratios are obtained for the 512\Theta512 "Lenna" image. Comparison of the systems with other results from the literature show that the fully adaptive system is quite competitive. 1. INTRODUCTION Multi-resolution decompositions using the discrete wavelet transform (DWT) have proven quite effective in the development of low-rate high-quality image coders. In [1], a discrete wavelet transform/entropy-constrained trellis coded quantization (DWT-ECTCQ) system was described that performed well against other subband and transform coders in the literature [2]--[7] at rates of 0...

Citations