Structure noise reduction and deconvolution of ultrasonic data using wavelet decomposition (ultrasonic flaw detection)

Abstract

The split-spectrum processing (SSP) technique is known to improve flaw detection in materials in which the coarse microstructure produces broadband noise of large amplitude, which masks useful signals. It is shown that the spectral decomposition used in the SSP is equivalent to the time-frequency Gabor decomposition. A generalized SSP based on a wavelet decomposition, in which the received signal is decomposed over a basis of elementary wavelets translated in frequency and dilated in time, and followed as in the conventional SSP by an optimization algorithm is introduced. The flexibility in the choice of the wavelet basis allows implementation of efficient signal decomposition. Three wavelet bases are presented: Gaussian-shaped wavelets, binary wavelets, and autoregressive wavelets. Applications of the SSP with wavelet decomposition to noise reduction and deconvolution are illustrated with experimental and simulated data.<>