基于DCT子空间失真测度的矢量量化VLSI结构设计

A VLSI Architecture for Vector Quantization Based on DCT Subspace Distortion Measure

本文提出了一种实现离散余弦变换（DCT）子空间失真测度矢量编码算法的VLSI结构。该结构利用DCT子空间映射，将失量量化失真测度维数从16降至4，并且结合最小覆盖搜索算法大幅度改善信噪比特性，从而在保证信噪比的条件下，使该编码器的实现规模降至一般矢量统码器的1／4。并且，本文对该算法作了进一步的改进，使实现更为简单。

A VLSI architeture for a new vector quantization encoding algorithm based on DCT subspace distortion measure is pressented. In the architecture,the dimensionality of the coder distorition measure is reduced form 16 to 4 by mapping the input vectors to DCT subspace, a searching method based on minimum covering set is also adopted to improve the performace of the encoding lgorithms inSNR.As a result,the implementation complexity of the encoder is reduced to 1/4 while the quality of the reconstructed image is maintained.