论文部分内容阅读
矢量量化(VQ)是近年来发展迅速的新型的高效信源编码技术,被广泛地应用于移动通信、卫星通信和保密通信等领域。而在矢量量化中,码本的优劣直接影响到 VQ 技术的性能。本文在研究了几种传统的建本算法的基础上,提出了一种高质量的量化方式——非均匀矢量量化(NUVO),并在此基础上提出了适应 NUVO 的建本算法——多级阈值聚类算法(Multi—Stage Threshold Clustering Algorithm——MSTC)。该方案主要建立在矢量自身的概率密度函数(pdf)之上,因此,在质量上有了较大的提高。
Vector quantization (VQ) is a new and efficient source coding technique developed rapidly in recent years. It is widely used in mobile communications, satellite communications and secure communications. In vector quantization, the merits of codebook directly affect the performance of VQ technology. Based on the study of several traditional algorithms, this paper proposes a high-quality quantization method called Non-uniform Vector Quantization (NUVO), and on the basis of this, proposes a built-in algorithm to adapt to NUVO Multi-Stage Threshold Clustering Algorithm - MSTC. The scheme is mainly based on the probability density function (pdf) of the vector itself, and therefore has a great improvement in quality.