多抽样率小波信号分解

Signal Decomposition Based on Multirate Wavelet Transform

在信号分析过程中，用二进小波对信号进行分解以便观察信号在不同子带内的时频特征是一种非常有效的时频分析方法。但由于受信号采样频率的限制，往往使所感兴趣的信号分量不能完整地出现在某一子带内。既使是提高小波分解层数或者用小波包技术也不能很好地解决这一问题。文章提出多抽样率小波信号分解方法，该方法根据目标信号的频带宽度，通过对原始信号的插值和抽取改变抽样频率，以实现不同子带分量的分离。

In signal analysis, the dyadic wavelet is always used for signal decomposition, which is an effective time-frequency analysis method for observing the time-frequency property of subband signals. But sometimes, the component of signal that authors interested in does not appear completely in one subband. Whether authors adopt the method of increasing the layers of wavelet decomposition or the technique of wavelet-packet, this problem can not be resolved ideally. In this paper, a new method of wavelet signal decomposition is proposed. According to the spectrum distribution of the destination signal, this method adjust the sampling rate in digital field by use of the technique of decimation and interpolation of digital signal so that different signal component can be separated.