基于多级SMVQ的图象编码算法

边缘匹配矢量量化器(SMVQ)是有限状态矢量量化器(FSVQ)的一个分支。该量化器适合于对图像块间相关性高的图像进行压缩编码,其优点是在比特率相近的情况下,编码质量高于传统的穷尽搜索矢量量化编码器,但其缺点是计算量大和比特率固定。本文提出了一种改进的边缘匹配矢量量化器。测试结果表明,该算法是变比特率编码算法,它比边缘匹配矢量量化器的比特率低,编码速度快,编码质量得到提高。

An Image Encoding Algorithm Based on the Multilevel Side- Match Vector Quantizer

The Side- Match Vector Quantizer(SMVQ) is a branchof the Finite- State Vector Quantizer (FSVQ). This quantizer issuitable for encoding the image in which there is a high correla-tion between neighboring image blocks. The virtue of this quan-tizer is that it can obtain higher image encoding quality than thetraditional full- search vector quantizer with nearly the same bitrate. However, it has the shortcomings of complex computationand fixed bit rate. A modified side- match vector quantizer ispresented in this paper. The experimental result shows that theproposed vector quantizer is bit- rate variable. Compared withthe SMVQ, the bit rate of the proposed algorithm is lower, itsencoding time is shorter, and its encoding quality is a little bet-ter.