在查阅和分析多级矢量量化和模拟退火技术有关文献资料的基础上,阐述了矢量量化最优码书的形成条件,并以多级矢量量化和模拟退火技术为基础,提出了一种基于模拟退火技术的多级矢量量化编码方案,该方案充分弥补了多级矢量量化和模拟退火技术在图象编码中应用的各自不足,并且发挥了多级矢量量化和模拟退火技术在图象编码中应用的各自优点,理论和实验都证明,该算法不仅能降低计算复杂度和码书存储量,而且图象恢复效果较好.

In this paper, First papers about multistage vector quantization (MSVQ) and Simulated annealing (SA) technique were consulted, based on a multistage vector quantization and Simulated annealing technique, we researched optimal codebook conditions of VQ, and present the algorithm of multistage vector quantization based on simulated annealing(SA-MSVQ). The scheme give full play to MSVQ and SA' s merits for image coding, counteract own weakness of multistage vector quantization based on simulated annealing. Theory and experimental results show that not only computation and storage is significantly lower, but also obtained a good image resume under the high compress ratio.