图象编码已经成为当今计算机世界的重要问题,而矢量量化(VQ)又是近年来有损图象压缩的一种重要技术,它的优点是比特率低以及解码简单,但是其穷尽搜索编码计算量较大.为了降低编码时间,已经有多种快速算法出现在一些文献中,然而这些算法往往不能进一步降低比特率.为了解决这一问题,因此提出了一种新颖的快速相关矢量量化(CVQ)图象编码算法.该算法对图象块的编码采用对角顺序,即在编码过程中根据当前图象块(输入矢量)与已编码的邻块之间的相关性来预测输入矢量的编码标号,从而大大降低了每个输入矢量平均码字的搜索范围

Image coding has been the important problem in the computer world nowadays. Vector quantization (VQ) is a lossy image compression technique presented recently. It has the advantages of low bit rate and simple decoding method. However,the encoding phase of the full search method need much computation. In order to reduce the encoding time,a lot of fast search algorithms are presented in the literatures. However,lots of these methods cannot further reduce the bit rate. In this paper,a novel image encoding algorithm based on the fast correlation vector quantization (CVQ) is presented. The diagonal encoding sequence is adopted in this paper. During the encoding process,the correlation between the current processing block and the adjacent encoded blocks is used to predict the index of the input vector,thus both the average codeword searching range of each input vector and the bit rate are greatly reduced. Test results show that,compared with the image encoding algorithms based on the conventional full-search method,the partial distortion search method and the double test method,the proposed algorithm needs much shorter encoding time and lower bit rate,although the encoding quality of the proposed algorithm is a little degraded.