A new digital image compression algorithm, HADZTRVQ, is presented based on wavelet transform and vector quantization. It embeds the concept of variable vector dimension VQ into our previously developed method WRDADRVQ, which is also based upon wavelet VQ. In the new algorithm, rate-distortion optimization is used to determine the optimal vector dimensions, number of coding units, and zero-tree strategies. By applying multiple vector dimensions, HADZTRVQ is supposed to outperform the previous algorithms of WRDADRVQ. The performances of these two algorithms as well as SPIHT are compared. The results show that HADZTRVQ significantly outperforms WRDADRVQ, but has not achieved a performance comparable with SPIHT. |