论文部分内容阅读
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 between the subbands generated by wavelet transform. However, in the current vector quantization based on wavelet transform, the code vectors often consist of inband coefficients and do not fully utilize the interband correlation. In this paper, a new cross-band vector quantization algorithm is proposed. The code vector consists of the corresponding coefficients in the same sub-band. Under the three-dimensional wavelet transform, the first-order sub-band code vector dimension is 48D, the second level is 12D, the third level is 3D. The code vector thus formed not only matches the importance of sub-bands at all levels, but also utilizes in-band and inter-band correlation. At the same bit rate, the signal to noise ratio PSNR of the reconstructed image is smaller than the in-band The traditional in-band VQ consisting of coefficients is 1 to 2 dB higher.