论文部分内容阅读
本文提出一种矢量压缩编码新方法—线上投影法 .它将输入矢量用它在某条空间直线上的投影近似表示 ,而用决定这条直线的两个参考点的序号和一个反映该投影点相对于两参考点位置的比例因子作为编码 .由于一个大小为N的矢量量化码书中的码字可以确定N(N - 1) / 2条直线 ,因此这种方法可用较小的码书获得很高的编码精度 .理论分析和实验结果表明 :码书大小为N的线上投影法的编码精度与码书大小为N2 的矢量量化法相当 ,并且明显优于用两个大小为N的码书构成的两级矢量量化法 ,而其码书生成和编码过程的计算复杂度均远远低于后者 .它将是矢量信号高精度压缩编码的一种强有力的手段 .
In this paper, we propose a new method of vector compression coding, the online projection method, which approximates the input vector with its projection on a straight line in a space and uses the numbers of the two reference points that determine the line and a projection that reflects the projection Since the code word in a vector quantization codebook of size N can determine N (N - 1) / 2 straight lines, this method can use a smaller codebook And obtain high coding precision.Theoretical analysis and experimental results show that the coding accuracy of the line projection method with codebook size N is equivalent to the vector quantization method with codebook size N2 and is obviously superior to that with two sizes of N The two-level vector quantization method, which is composed of codebook and codebook, has much lower computational complexity than the latter, which will be a powerful means of high precision compression coding of vector signal.