EUVIMEDEUVIMED entdecken
Crossref

HiFC-GAN: Hierarchical Feature-Constrained GAN for Optical-to-SAR Transfer in SAR Target Classification

Hao Zheng, Meiguang Zheng, Zhigang Hu, Liu Yang, Aikun Xu, Tingxuan Chen, Rongchang Zhao, Boyu Wang

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (16) · 13387-13395

DOI: 10.1609/aaai.v40i16.38342 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Hao Zheng und 7 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The limited availability of high-quality training data poses a persistent challenge for synthetic aperture radar (SAR) target classification. Existing data augmentation methods mainly adopt a simplistic application of GAN-based style transfer techniques to directly synthesize pseudo-SAR images from optical images. However, our in-depth analysis of this cross-modal conversion reveals that such straightforward strategies primarily focus on transferring high-level semantic information (e.g., target shapes), thus failing to adequately capture the essential low-level features unique to SAR imagery (e.g., scattering textures). To address this inherent trade-off between high-level semantic preservation and low-level feature authenticity, we propose a Hierarchical Feature-Constrained GAN (HiFC-GAN) tailored for optical-to-SAR style transfer. Specifically, HiFC-GAN enhances the representation of low-level SAR features by introducing local texture contrast constraints at shallow layers, while introducing explicit feature mapping constraints at deeper layers to maintain high-level semantic consistency throughout the reconstruction process. Experimental results demonstrate that HiFC-GAN significantly outperforms existing GAN-based techniques in image generation quality, particularly improving the low-level feature authenticity of pseudo-SAR images. Moreover, the generated pseudo-SAR images further improve the performance of downstream target classification tasks, yielding accuracy gains ranging from 3.56% to 5.90% on average with mainstream CNN-based models.

Zitieren

APA

Zheng, H., Zheng, M., Hu, Z., Yang, L., Xu, A., Chen, T., Zhao, R., & Wang, B. (2026). HiFC-GAN: Hierarchical Feature-Constrained GAN for Optical-to-SAR Transfer in SAR Target Classification. Proceedings of the AAAI Conference on Artificial Intelligence, 40(16), 13387-13395. https://doi.org/10.1609/aaai.v40i16.38342

Vancouver

Zheng H, Zheng M, Hu Z, Yang L, Xu A, Chen T, et al. HiFC-GAN: Hierarchical Feature-Constrained GAN for Optical-to-SAR Transfer in SAR Target Classification. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(16):13387-13395. doi:10.1609/aaai.v40i16.38342

BibTeX
@article{Zheng2026,
  author = {Zheng, Hao and Zheng, Meiguang and Hu, Zhigang and Yang, Liu and Xu, Aikun and Chen, Tingxuan and Zhao, Rongchang and Wang, Boyu},
  title = {HiFC-GAN: Hierarchical Feature-Constrained GAN for Optical-to-SAR Transfer in SAR Target Classification},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {16},
  pages = {13387--13395},
  doi = {10.1609/aaai.v40i16.38342}
}
Kostenlos starten →