基于双正交小波的快速矢量量化算法

提出了一种用于图象压缩的矢量量化算法．该算法通过构造符合图象小波变换系数特征的跨频带矢量，利用小波系数之间的相关性，提高了图象的编码效率和重构质量．同时，该算法又采用了两种新的矢量量化技术——非线性插补矢量量化（ＮＬＩＶＱ）和渐进构造聚类（ＰＣＣ），提高了矢量量化的速度和码书质量．实验结果证明，该算法在比特率为０．１７２ｂｐｐ的条件下仍能获得ＰＳＮＲ＞３４ｄＢ的高质量重构图象．研究图象小波变换系数的固有特性是提高矢量量化性能的关键，而提高矢量量化速度是这类算法得以实用的重要前提．

Biorthogonal Wavelet Based Fast Vector Quantization Algorithm

A novel wavelet based vector quantization(VQ) algorithm is presented. The correlation of the wavelet coefficients is fully removed through constructing the band cross vector; therefore, the high coding efficiency and reconstructed image quality are both obtained. In addition, two new VQ technologies are used in this algorithm to improve the coding efficiency and the code book quality. The simulation results show that this coding scheme can get a very high reconstruction quality (PSNR>34 dB) at a very low bit rate(0.172 bpp). A new idea is presented to find a way to combine the wavelet transform and VQ together more efficiently. Through the experiment, it can be seen that the property of the transformed image coefficients is the key to improve VQ performance, and the speed of VQ is an important factor for real time application of this kind of image coding algorithms.