本文讨论了矢量量化的速率与失真,证明了对于给定的K维输出码矢数N,以最小可达失真Dmin进行信源编码时,最小编码速率是码书大小为N的K维最佳矢量量化器的输出熵;对于具有对称分布的码书,证明了矢量量化器的输出熵必不小于每矢量1比特;最后,导出了反映矢量量化器工作特性和率失真函数之间的数量关系的不等式串。 This paper discusses the rate and distortion of vector quantization, and proves that for a given K of output code vector N, the minimum coding rate is the K-dimensional best codebook size N when the source coding is performed with the minimum reachable distortion Dmin The output entropy of the vector quantizer; for a codebook with symmetric distribution, it is proved that the output entropy of the vector quantizer must be no less than 1 bit per vector; finally, the quantitative relationship between the performance of the vector quantizer and the rate-distortion function Inequality string.