Ultrasonic wave field image augmentation in PZT sensors using generative machine learning and Coulomb coupling
Open Access
- 20 February 2025
- journal article
- research article
- Published by EDP Sciences in Acta Acustica
- Vol. 9, 15
Abstract
This paper presents an approach to overcome the time-intensive nature of the Coulomb coupling imaging method by employing Generative Adversarial Networks (GANs) for data augmentation. Coulomb coupling, an experimental technique, is essential for visualizing ultrasonic wave propagation in piezoelectric materials and is valuable in various domains including materials research. It provides valuable insights such as finding mechanical properties and detecting anomalies in piezoelectric materials. However, the efficiency of this method is hindered by traditional time expansive point-by-point scanning. Integrating advanced machine learning into Coulomb coupling imaging has emerged as a promising solution to address this issue. Nonetheless, the lack of sufficient data has been a significant challenge. The key contribution is the use of GANs to create synthetic yet realistic images from a limited set of real data, effectively overcoming the issue of data scarcity. A large number of artificial images were successfully generated, expediting model training and enhancing generalization. This study is the first to use GANs in Coulomb coupling imaging, showing its transformative potential. By overcoming data limitations, the proposed approach enhances Coulomb coupling imaging and enables its integration with advanced technologies like AI-driven predictive modeling and real-time adaptive imaging. This opens new frontiers for applications in materials science and other imaging modalities.Keywords
Funding Information
- NA (NA)
This publication has 33 references indexed in Scilit:
- Data augmentation for medical imaging: A systematic literature reviewComputers in Biology and Medicine, 2022
- Ultrasonic image denoising using machine learning in point contact excitation and detection methodUltrasonics, 2022
- Synthetic Data generation using DCGAN for improved traffic sign recognitionNeural Computing and Applications, 2021
- Damage localization in piezo-ceramic using ultrasonic waves excited by dual point contact excitation and detection schemeUltrasonics, 2020
- Addressing Neural Network Robustness with Mixup and Targeted Labeling Adversarial Training2020
- GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classificationNeurocomputing, 2018
- Diverse Image-to-Image Translation via Disentangled Representations2018
- Conditional Infilling GANs for Data Augmentation in Mammogram Classification2018
- Modeling of Coulomb coupling and acoustic wave propagation in LiNbO3Ultrasonics, 2008
- A Decision-Theoretic Generalization of On-Line Learning and an Application to BoostingJournal of Computer and System Sciences, 1997