基于二叉树搜索的小波变换信号压缩

正交离散小波变换由于缺乏平移不变性,会对信号压缩的性能造成很大的影响。为避免这一缺陷,该文在每个变换尺度上均对偶采样和奇采样的结果进行保留,产生一个三叉树结构,由此二叉树的任意一路径均可重构原信号,按照一定的准则可以确定最优分解路线。仿真结果表明,该变换方式较标准正交离散小波变换有更高的压缩性能。

Signal compression by bitree-search-based wavelet transform

ODWT (Orthogonal Discrete Wavelet Transform) is translation-variant, which will depress the compression efficiency. To avoid the deficiency, by holding all the even and odd indexed subsamples in every scale, a bitree structure is formed here. From any route of the bitree the signal can be reconstructed. According to sonic given criterion, a best transform route can be determined. The simulations show that the compression performance of the best transform route is better than that of the ordinary ODWT.