An automated framework for the extraction of ultrasonic echoes embedded in noise

• Published in: Proceedings of Meetings on Acoustics Vol. 19; no. 1
• Main Authors: Pedrycz, Adam, Valero, Henri-Pierre, Hori, Hiroshi, Nishimiya, Kojiro, Sugiyama, Hitoshi, Sakata, Yoshino
• Format: Conference Proceeding
• Language: English
• Published: 02.06.2013
• ISSN: 1939-800X
• DOI: 10.1121/1.4800843

Proposed is an automated framework for the extraction and characterization of the arriving echo in ultrasonic signals embedded in high noise. Commonly, in order to correctly characterize the first echo hidden within a noise-ridden signal, multiple traces are stacked in a gather to improve the SNR, hence facilitating easier extraction and characterization of the recorded echo. Such first order statistical methods require multiple traces and usually fall short in the accuracy of the echo estimate when the variance of the noise does not belong to a known distribution. To mitigate this problem, a framework has been developed comprised of a multi-step procedure, i.e., pre-processing, localization, gating and finally parameterization of the given echo. This automatic framework operates on single traces and does not require the setting of processing parameters. By means of this method, the true echo can be extracted in one-shot from other overlapping noise components. Furthermore, because the method operates on a trace-by-trace basis, it is insensitive to large non-stationarities in the baseline. Experiments conducted using aluminum reflector pulse-echo laboratory data demonstrate the effective extraction of the true echo under the presence of noise, and its implications in regard to ultrasonic imaging.