一种基于小波变换的跨带矢量量化  

Crossband vector quantization based on wavelet transform

作　　者：王光学[1,2] 孙光宇[1,2] 曹长修[1,2] 刘一武 

机构地区：[1]重庆大学电信学院 [2]贵州教育学院

出　　处：《光学技术》1999年第2期45-47,共3页Optical Technique

摘　　要：由小波变换产生的子带之间存在着较强的相关性，然而目前的在基于小波变换的图像编码矢量量化中，码矢往往仅由带内系数组成，没有充分利用带间相关性。本文提出一种新的跨带矢量量化算法，码矢由在同级子带中的相应系数组成。在三次小波变换下，第一级子带的码矢维数为４８Ｄ，第二级为１２Ｄ，第三级为３Ｄ。这样形成的码矢既匹配了各级子带的重要性，又利用到了带内和带间相关性，在相同的比特率下，其重构图像的信噪比ＰＳＮＲ比码矢仅由带内系数组成的传统带内ＶＱ高出１～２ｄＢ。There is a strong correlation among sub bands from wavelet transform. However, in the present image coding based on wavelet transform, codevectors are often composed of only intraband coefficients, without effectively utilizing interband correlation. In this paper, a new algorithm of crossband vector quantization is presented, in which codevectors consist of related coefficients of same level crossband. Under three stages wavelet transform, the dimension of codevetors of the first level crossband is 48D, the second level is 12D, the third is 3D. The codevectors constructed in that way not match the importance of each level subband, but also use the correlation of interband and interband. At a same rate PSNR of its reconstructed image is 1～2dB higher than the conventional intraband VQs where codevector is made of only intraband coefficients.

关 键 词：图像编码 矢量量化 小波变换 跨带 

分 类 号：O438[机械工程—光学工程] TP391.41[理学—光学]