摘要
由实际应用可得,作为一种新的多辨认分析方法的小波变换,由于具有多分辨特征和时频局部性,可同时进行频域和时域分析,所以特别适用于进行非平稳信号的处理。首先介绍了小波分析的信号去噪原理,其次接着阐述了4种去噪算法:小波分解与重构去噪方法、模极大值检测法、阈值法及平移不变量小波去噪法。对于叠加了高斯白噪声的仿真信号,分别将上述的4种方法用于去噪处理,并通过仿真过程不同及图形差异对几种方法进行比较。
From an engineering viewpoint,wavelet transform as a new multi-resolution analysis method,for its time-frequency localization and multi-resolution characteristics,and its time-domain and frequency domain analysis in the meantime,is particularly suitable to handling the non-stationary signals.This paper first describes the signal de-noising based on wavelet analysis theory,and then the four de-noising algorithms including the wavelet decomposition and reconstruction,the modulus maxima detection,the threshold value and translation invariant wavelet de-noising.For the simulated signal with the superposition of Gaussian white noise,these methods are respectively used to remove the Gaussian white noise from the simulated signal.And based on the difference of the simulation process and graph,these algorithms are compared and discussed.
出处
《通信技术》
2010年第9期79-81,84,共4页
Communications Technology
基金
国家863重点项目(批准号:2009AA01Z403)
国家863重点项目(批准号:2009AA01Z435)
关键词
小波变换
非平稳信号
信号去噪
高斯白噪声
仿真信号
wavelet transform
non-stationary signal
signal de-noising
gaussian white noise
simulated signal
