Discrete frequency slice wavelet transform

Zhonghong Yan

doi:10.1016/j.ymssp.2017.04.019

2017-11-01

PRE

DOI：10.1016/j.ymssp.2017.04.019

原文链接

原文求助

摘要

En 中文

This paper introduces a new kind of Time-Frequency Representation (TFR) method called Discrete Frequency Slice Wavelet Transform (DFSWT). It is an improved version of Frequency Slice Wavelet Transform (FSWT). The previous researches on FSWT show that it is a new efficient TFR in an easy way without strict limitation as traditional wavelet theory. DFSWT as Well as FSWT are defined directly in frequency domain, and still keep its properties in time-frequency domain as FSWT decomposition, reconstruction and filter design, etc. However, the original signal is decomposed and reconstructed on a Chosen Frequency Domains (CFD) as need of application. CFD means that the decomposition and reconstruction are not completed on all frequency components. At first, it is important to discuss the necessary condition of CFD to reconstruct the original signal. And then based on norm 12, an optimization algorithm is introduced to reconstruct the original signal even accurately. Finally, for a test example, the TFR analysis of a real life signal is shown. Some conclusions are drawn that the concept of CFD is very useful to application, and the DFSWT can become a simple and easy tool of TFR method, and also provide a new idea of low speed sampling of high frequency signal in applications. (C) 2017 Published by Elsevier Ltd.

Keyword:

Time-frequency analysis

Free style wavelet base

New method of signal processing

Bio-signal decomposition and reconstruction

Signal filter