Noise-Robust Incremental Target Recognition Using Residual Free-Energy Gating for Polarimetric Radar HRRP Sensors
IEEE Sensors Journal · 2026-08 · 26 (16) · 25006-25019
DOI: 10.1109/jsen.2026.3710693 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-08. Verfasst von Xiaoyu Xu und 3 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Gefördert von: 2026 Youth Fund of Naval Aviation University.
Abstract (Originalsprache)
This article addresses the electromagnetic interference, feature distortion, and catastrophic forgetting in polarimetric radar high-resolution range profile (HRRP) ship incremental recognition. A residual free-energy enhanced network (RFE-Net) is proposed for noise-adaptive incremental target recognition. A multiscale dual-residual convolutional module is designed to capture multiscale scattering characteristics of radar HRRP echoes. Combined with thermodynamic free-energy theory, an interpretable engineering free-energy model is established, and a free-energy gating (FEG) mechanism is developed to implement the adaptive noise suppression without fixed thresholds. Meanwhile, a polarization-aware weighted fusion strategy is adopted to mine complementary information between HH and VV dual-polarization channels. Moreover, a free-energy-aware knowledge distillation scheme is introduced to align deep feature and free-energy distributions across historical and novel tasks, which effectively alleviates catastrophic forgetting. Experiments are carried out on a measured 12-class ship HRRP dataset. The proposed method achieves an average incremental recognition accuracy of 96.53% and maintains satisfactory performance under Gaussian, impulsive, and hybrid noise at low-signal-to-noise ratio (SNR) levels. Ablation and comparative experiments verify the efficacy of each designed module and the overall performance of RFE-Net, which is applicable to ship recognition tasks in complex marine electromagnetic scenarios.
Förderung
- 2026 Youth Fund of Naval Aviation University
Zitieren
APA
Xu, X., Zhu, M., Dan, B., & Xu, L. (2026). Noise-Robust Incremental Target Recognition Using Residual Free-Energy Gating for Polarimetric Radar HRRP Sensors. IEEE Sensors Journal, 26(16), 25006-25019. https://doi.org/10.1109/jsen.2026.3710693
Vancouver
Xu X, Zhu M, Dan B, Xu L. Noise-Robust Incremental Target Recognition Using Residual Free-Energy Gating for Polarimetric Radar HRRP Sensors. IEEE Sensors Journal. 2026;26(16):25006-25019. doi:10.1109/jsen.2026.3710693
BibTeX
@article{Xu2026,
author = {Xu, Xiaoyu and Zhu, Mingbo and Dan, Bo and Xu, Le},
title = {Noise-Robust Incremental Target Recognition Using Residual Free-Energy Gating for Polarimetric Radar HRRP Sensors},
journal = {IEEE Sensors Journal},
year = {2026},
volume = {26},
number = {16},
pages = {25006--25019},
doi = {10.1109/jsen.2026.3710693}
}
EUVIMED