Flaw detection on highly scattering materials using multiresolution analysis with time-frequency thresholding

Abstract

In ultrasonic non-destructive evaluation of highly scattering materials, detection of flaw echoes is difficult due to the masking effects of the structural noise. This paper presents a method for the reduction of the structural noise on highly scattering aterials. The method consists of a multi-resolution analysis using wavelets and a time-frequency thresholding applied to the approximation coefficients. This threshold is estimated by considering the mean and the standard deviation of the time-frequency energy representation. This method exploits spectral properties of the structural noise and the incoming flaw signal, resulting in a processed signal that maintains the defect pulse shape. In order to validate this strategy, synthetic and experimental signals have been evaluated. The performance of this method is compared with known selection rules as Minimax and Universal threshold.