自适应补偿矢量量化

Image Vector Quantization With Adaptive Compensation

提出了一种基于ＬＢＧ码书设计的新的图像矢量量化算法。该算法利用图像信号在正交矢量空间中的能量集中性，有效地减小了码书的搜索范围，加快了矢量量化速度。同时利用原始图像和重建图像之间的差值进行了自适应补偿，在保证较高压缩比的同时，有效地克服了矢量量化的致命缺陷，即重建图像存在严重的方块效应。

In this paper,a new image vector quantization algorithm is presented. This algorithm utilizes the compactness property of signal energy on transform domain to reduce the searching range of codebook and speed up the quantization. The defference values between the input image and reconstructed image are used to compensate adaptively. Experiment results show that this algorithm can overcome the block artifacts which is inherently in conventional VQ as well as keep high compression ratio.