Using wavelet transform (WT) for increasing signal-to-noise ratio (SNR) of discrete-time signals corrupted by additive noise is explained and compared with some other techniques (averaging, frequency filtration, correlation). Signal processing for de-noising is applied to basic periodical signals and repeated transients (in nondestructive ultrasonic testing of welds, where presence of flaws should be detected). Results of both computer simulations and measurements are reported, and some best suitable wavelets, levels of signal decomposition and methods and parameters of thresholding are given. A new efficient method of wavelet thresholding suitable for ultrasonic flaws detection in welds testing is described as a part of practical wavelets SNR enhancement (SNRE) application, and correlation function used for the same purpose is also described.. Wavelet Toolbox of MATLAB environment is used both for computer simulations and practical signal de-noising.
[1]
B. Widrow,et al.
Adaptive noise cancelling: Principles and applications
,
1975
.
[2]
Gérard Faucon,et al.
Using the coherence function for noise reduction
,
1992
.
[3]
David L. Donoho,et al.
De-noising by soft-thresholding
,
1995,
IEEE Trans. Inf. Theory.
[4]
Yih-Chyun Jenq,et al.
Discrete-time method for signal-to-noise power ratio measurement
,
1996
.
[5]
Truong Q. Nguyen,et al.
Wavelets and filter banks
,
1996
.
[6]
Alfred Mertins,et al.
Signal Analysis: Wavelets, Filter Banks, Time-Frequency Transforms and Applications
,
1999
.
[7]
M. Kreidl,et al.
Reducing Ultrasonic Signal Noise by Algorithms based on Wavelet Thresholding
,
2002
.