Session: Research Posters

Paper Number: 119400

119400 - Apex-Shifted Radon Transform-Based Direct Arrival Removal for Ultrasonic Array Measurements 

We propose an apex-shifted Radon transform (ASRT)-based technique to mitigate surface wave artifacts (SWAs) in the total focusing method (TFM) for ultrasonic imaging. We also demonstrate the ASRT can be used to isolate a wave packet from surface waves in the presence of a temperature change, which is difficult to perform using baseline subtraction for structural health monitoring purposes. SWAs are a challenge in reflection mode scans, occurring when direct P-waves and surface waves (collectively referred to here as direct arrivals) induce strong near-surface artifacts in the TFM images generated from the scan. This interference results in a “dead-zone” near the array that is difficult to image due to SWA crosstalk, and the artifacts lower the overall visual clarity of TFM reconstructions, potentially obscuring damage sites or other regions of interest. To cope with this effect, signals from this SWA dead-zone region are commonly deleted, a simple compensation strategy that reliably diminishes the artifacts but also removes the signature of any potential damage or regions of interest in the dead-zone region. To overcome the issue of SWAs while enabling imaging of the dead-zone, an ASRT-based image processing algorithm is introduced to selectively mute the direct arrivals in full matrix capture data used for TFM. The ASRT is a mathematical transform common in signal processing applications like seismology, and is the core of the proposed technique. It targets specific geometry from time-space domain full matrix capture (FMC) data via a line-integral transform parameterized by curvature, and compresses the corresponding wave packets into point-like regions in the Radon domain. In this new domain, overlapping events in the original domain are represented as sparse discrete regions where undesirable information can be muted without destroying nearby data. After an ASRT-based mute process is applied, the signature of the direct arrivals is greatly reduced in the time-space domain, and the SWAs are likewise heavily mitigated.

In this presentation, we begin with a definition and illustration of SWAs, and how some scanning setups are particularly prone to such artifacts. Next, an explanation of how ASRT can remove direct arrivals in FMC data is provided, and the method is explained in detail including some best practices for performing an ASRT mute. Then, the proposed SWA-mute technique is demonstrated on data generated from spectral element wave propagation simulations and performs comparably to the baseline subtraction approach, which is the ideal case of subtracting the signals of interest from an empty domain of the same size and material property distribution. For verification, the results are validated on the TFM images of experimental ultrasonic measurements polluted by SWAs, and the performance gap between synthetic and experimental results is discussed. Finally, we provide some comments about the effectiveness of the technique and future work is described. The ASRT-based SWA mute process is concluded to show promising results in both the numerical and experimental cases.

Presenting Author: Augustine Loshelder The University of Alabama

Presenting Author Biography: Augustine (Gus) Loshelder is a second-year PhD student at the University of Alabama's Lab for Computational Imaging and Smart Structures (CISS), where he has been active since late 2020. In his research career, he has been published in the journal Structural Health Monitoring and his work has been presented in the American Society for Nondestructive Testing Research Symposium 2022. He has received several departmental awards including outstanding research assistant as both and undergraduate and Master's student. His research focuses include ultrasound signal processing techniques, nondestructive evaluation, deep learning, and physics-informed machine learning.

Authors:

Augustine Loshelder The University of Alabama
Jiaze He The University of Alabama
John Day The University of Alabama