Itakura–Saito Divergence-Based Physics-Guided Source Separation for Electromagnetic Near- and Remote-Field Sensing
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 6005413-6005413
DOI: 10.1109/tim.2026.3716460 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Zhaohu Yu und 3 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Gefördert von: National Key Research and Development Program of China, Horizontal Research Project through EXATEC Private Ltd.
Abstract (Originalsprache)
During electromagnetic-field diffusion in ferromagnetic structures, near- and remote-field regimes exhibit distinct spectral–spatial characteristics. However, defect-related components are often entangled with primary-field responses, magnetization effects, and nuisance backgrounds, which make physics-guided source separation essential for reliable electromagnetic sensing. Nonnegative matrix factorization (NMF) with Itakura–Saito (IS) divergence provides a scale-invariant decomposition framework and has the potential to bridge data-driven factorization and physically meaningful component separation. This study proposes a semisupervised physics-guided source separation framework for electromagnetic near-/remote-field sensing. The proposed method incorporates physical spectral priors to distinguish high-frequency near-field responses from low-frequency remote-field responses. Spatial smoothness constraints consistent with remote-field propagation and morphological templates derived from simulations or defect References are further introduced to guide the separation process. These priors guide the factorization toward physically meaningful bases and reduce ambiguous component mixing. Experiments testing on both artificial steel defects and natural defects verify that the proposed method can separate near-/remote-field responses and enhances defect-related activation maps with improved SNR and visualization clarity. This work provides a data-driven yet physically interpretable framework for electromagnetic source separation in near-/remote-field sensing.
Förderung
- National Key Research and Development Program of China
- Horizontal Research Project through EXATEC Private Ltd
Zitieren
APA
Yu, Z., Gao, B., Tian, G., & Ong, D. (2026). Itakura–Saito Divergence-Based Physics-Guided Source Separation for Electromagnetic Near- and Remote-Field Sensing. IEEE Transactions on Instrumentation and Measurement, 75, 6005413-6005413. https://doi.org/10.1109/tim.2026.3716460
Vancouver
Yu Z, Gao B, Tian G, Ong D. Itakura–Saito Divergence-Based Physics-Guided Source Separation for Electromagnetic Near- and Remote-Field Sensing. IEEE Transactions on Instrumentation and Measurement. 2026;75:6005413-6005413. doi:10.1109/tim.2026.3716460
BibTeX
@article{Yu2026,
author = {Yu, Zhaohu and Gao, Bin and Tian, Guiyun and Ong, Derek},
title = {Itakura–Saito Divergence-Based Physics-Guided Source Separation for Electromagnetic Near- and Remote-Field Sensing},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {6005413--6005413},
doi = {10.1109/tim.2026.3716460}
}
EUVIMED