ECHO: Estimating Continuous Radar Cross Section via Hierarchical Features of Objects
IEEE Transactions on Antennas and Propagation · 2026-08 · 74 (8) · 7780-7791
DOI: 10.1109/tap.2026.3686018 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Antennas and Propagation, 2026-08. Verfasst von Zehao Tang und 5 weiteren Autor:innen. Die Arbeit nennt 41 Literaturangaben. Gefördert von: National Key Research and Development Program of China.
Abstract (Originalsprache)
The radar cross section (RCS) is a fundamental electromagnetic (EM) property that characterizes how an object scatters incident waves. Traditional numerical methods for computing the RCS of large-scale objects often demand significant computational resources and time. In recent years, learning-based approaches have shown promise in near-field EM problems, yet their potential for accurate and efficient RCS estimation remains underexplored. Most existing methods adopt fully convolutional networks (FCNs), which tend to rely on global object features and struggle to capture fine-grained surface details or model the interaction between waves and fields effectively. To address these limitations, we propose estimating continuous radar cross section via hierarchical features of objects (ECHOs), a neural network framework for broadband, full-angle monostatic RCS estimation from 3-D input. Unlike FCN-based models, ECHO employs a coordinate-based multilayer perceptron as the decoder to regress the RCS at arbitrary incident angles, enabling continuous-angle estimation rather than discrete, angle-fixed prediction. To enhance the angular specificity, we design an angle querier module that extracts localized object features guided by the incident direction. Moreover, we introduce a wave encoder to effectively model variations across frequency and angle. Experiments show that ECHO achieves superior accuracy compared to FCN-based baselines while using 25% of their parameters. We further demonstrate its fast inference capability and validate the effectiveness of each module through ablation studies.
Förderung
- National Key Research and Development Program of China
Zitieren
APA
Tang, Z., Wu, K., Luo, Y., Wang, H., Zhang, S., & Leng, B. (2026). ECHO: Estimating Continuous Radar Cross Section via Hierarchical Features of Objects. IEEE Transactions on Antennas and Propagation, 74(8), 7780-7791. https://doi.org/10.1109/tap.2026.3686018
Vancouver
Tang Z, Wu K, Luo Y, Wang H, Zhang S, Leng B. ECHO: Estimating Continuous Radar Cross Section via Hierarchical Features of Objects. IEEE Transactions on Antennas and Propagation. 2026;74(8):7780-7791. doi:10.1109/tap.2026.3686018
BibTeX
@article{Tang2026,
author = {Tang, Zehao and Wu, Kaiming and Luo, Yuxuan and Wang, Hao and Zhang, Shuo and Leng, Biao},
title = {ECHO: Estimating Continuous Radar Cross Section via Hierarchical Features of Objects},
journal = {IEEE Transactions on Antennas and Propagation},
year = {2026},
volume = {74},
number = {8},
pages = {7780--7791},
doi = {10.1109/tap.2026.3686018}
}
EUVIMED