San Emeterio J., Pardo E., Rodriguez M., Ramos A.

Different digital signal processing techniques have been used for denoising pulse-echo traces in ultrasonic imaging / detection applications. A particular type of noise, which is usually called grain noise or structural noise, plays an important role in ultrasonic signal detection applications. This type of noise, originated from the addition of multiple echoes produced by randomly located scatters (grain boundaries) inside the inspected material, has a frequency band very similar to that of the echoes issuing from the defects or discontinuities to be detected. Conventional averaging and/or filtering techniques are not useful for grain noise reduction. Specific denoising methods have been proposed for processing these signals, based on either spatial diversity or frequency diversity, including Split Spectrum Processing and time-frequency-scale processing [1]. Wavelet processing offers great flexibility and is a well established technique for removing noise from signals. The usual discrimination between signal and noise consists of a thresholding and/or pruning of the coefficients in the transformed wavelet domain [2]. In this work, decomposition level dependent thresholds have been used to denoise ultrasonic signals. Thresholds for each decomposition level are estimated from their wavelet coefficients utilizing Universal, Minimax and SURE threshold selection rules. Discrete wavelet transform (DWT) and translation-invariant wavelet transforms are applied. Two different undecimated wavelet transform (UWT) processors, which we have specifically developed for noise reduction purposes, have been used. The efficiency in noise reduction, for single echo detection, is evaluated by means of the signal-to-noise ratio SNR enhancement using synthetic and experimental pulse-echo ultrasonic traces. [1] M.A. Rodríguez, J.L. San Emeterio, J.C. Lázaro, A. Ramos, Ultrasonics, Vol 42, pp. 847-851, 2004. [2] J.C. Lazaro, J.L. San Emeterio, A. Ramos, J.L. Fernandez, Ultrasonics, Vol 40, pp. 263-267, 2002.