本文提出了一种改进的图像矢量量化编码算法．该算法通过采用增加预测块，以及根据相邻块状态进行控制信息的条件熵编码等改进方法，进一步提高了算法的编码效率．为了便于算法实现，本文对图像的边缘引进了一致的处理方法．测试结果表明，相对于无记忆矢量量化编码算法，比特率约下降４０％，相对于文献［５］算法，比特率约下降１５％． This paper presents an improved image vector quantization coding algorithm. The algorithm further improves the coding efficiency of the algorithm by adopting improved methods such as adding prediction blocks and condition entropy coding of control information according to the status of neighboring blocks. In order to facilitate the realization of the algorithm, this paper introduced a consistent approach to the edge of the image. The test results show that the bit rate is about 40% lower than that of the memoryless vector quantization coding algorithm, and the bit rate is reduced by about 15% compared with the algorithm of [5].