基于多球体空间拓扑约束的雷达目标HRRP少样本开集识别方法

徐寒铮; 刘峥; 许述文; 郭泽坤

doi:10.12263/DZXB.20250705

您当前的位置：

首页 >

文章列表页 >

基于多球体空间拓扑约束的雷达目标HRRP少样本开集识别方法

学术论文 | 更新时间：2026-06-16

- 基于多球体空间拓扑约束的雷达目标HRRP少样本开集识别方法
- A Few-Shot Open-Set Recognition Method for Radar Target HRRP Based on Multi-Hypersphere Spatial Topological Constraints
- 电子学报 2026年54卷第3期页码：1209-1220
- 作者机构：
  
  西安电子科技大学雷达信号处理全国重点实验室，陕西西安 710071
- 作者简介：
  
  徐寒铮男，2000年1月出生于福建省宁德市。现为西安电子科技大学博士研究生。主要研究方向为雷达自动目标识别、深度学习、模式识别和小样本学习。中国电子学会会员编号：E190185703A。 E-mail: 24021110921@stu.xidian.edu.cn
  刘峥男，1964年3月出生于陕西省咸阳市。现为西安电子科技大学雷达信号处理全国重点实验室教授、博士生导师。主要研究方向为雷达信号处理的理论与系统设计、雷达精确制导技术、多传感器协同探测与信息融合等。 E-mail: lz@xidian.edu.cn
  许述文男，1985年11月出生于安徽省黄山市。现为西安电子科技大学雷达信号处理全国重点实验室副主任、教授、博士生导师。主要研究方向为雷达目标检测、机器学习、时频分析和SAR图像处理。中国电子学会会员编号：E190012872S。 E-mail: swxu@mail.xidian.edu.cn
  郭泽坤男，1994年8月出生于陕西省咸阳市。现为西安电子科技大学博士研究生。主要研究方向为雷达目标识别、深度学习、小样本学习。 E-mail: zkguo@xidian.edu.cn
- 基金信息：
  
  国家自然科学基金(62371382;U24A20217)
- DOI：10.12263/DZXB.20250705
  中图分类号： TN959.1+7;
- 收稿：2025-08-14，
  
  录用：2026-03-04，
  
  纸质出版：2026-03-25
- 稿件说明：
移动端阅览
徐寒铮, 刘峥, 许述文, 等. 基于多球体空间拓扑约束的雷达目标HRRP少样本开集识别方法[J]. 电子学报, 2026, 54(03): 1209-1220.

XU Hanzheng, LIU Zheng, XU Shuwen, et al. A Few-Shot Open-Set Recognition Method for Radar Target HRRP Based on Multi-Hypersphere Spatial Topological Constraints[J]. Acta Electronica Sinica, 2026, 54(03): 1209-1220.
徐寒铮, 刘峥, 许述文, 等. 基于多球体空间拓扑约束的雷达目标HRRP少样本开集识别方法[J]. 电子学报, 2026, 54(03): 1209-1220. DOI：10.12263/DZXB.20250705

XU Hanzheng, LIU Zheng, XU Shuwen, et al. A Few-Shot Open-Set Recognition Method for Radar Target HRRP Based on Multi-Hypersphere Spatial Topological Constraints[J]. Acta Electronica Sinica, 2026, 54(03): 1209-1220. DOI：10.12263/DZXB.20250705

摘要

雷达目标高分辨距离像（High Resolution Range Profile，HRRP）因其能够有效表征目标的几何结构与电磁散射特性，在雷达目标识别领域具有重要的应用价值。近年来，深度学习方法凭借其强大的特征表征能力被广泛应用于HRRP目标识别任务。然而，现有基于深度学习的HRRP目标识别方法在少样本开集识别（Few-Shot Open-Set Recognition，FSOSR）场景中，其性能因训练样本少且本身不具备对未知类别的判别能力而受限。为此，本文提出一种基于多球体空间拓扑约束的雷达目标HRRP FSOSR方法。首先，在训练策略上引入元学习框架，从类别充足、样本丰富的辅助数据集采样大量相似且互不重复的任务作为训练单元，学习跨任务共通的知识，使模型在样本稀缺的任务数据集上具备快速泛化能力，缓解少样本模型过拟合问题；其次，在此基础上，设计适用于少样本场景的多球体决策边界建模机制，以超球体分别建模各已知类特征子空间，形成由多球体组成的已知类空间分布，从而隐式建模未知类分布；再次，在多球体之间引入自适应间隔，维持已知类间特定空间拓扑关系，提升决策边界的精细度和鲁棒性；最后，提出难分样本学习策略，先经难分挖掘筛选高价值样本，再借助样本对加权机制定量刻画样本难分程度，使模型依据权重进行针对性学习，从而利用高价值难分样本挖掘更精细的差异化特征，以增强模型对细粒度未知类判别能力。实验结果表明：该方法在5-shot及10-shot场景下，相较于现有方法，准确率分别提升6.17个百分点和2.94个百分点，AUROC指标分别提升13.1个百分点和12.94个百分点，验证了该方法的有效性与稳健性。此外，通过在瑞芯微RK3588嵌入式人工智能（Artificial Intelligence，AI）芯片上完成模型部署，所提方法推理时延2.197 ms，功耗为2.25 W，充分验证了该方法的工程可实现性。该方法适用于未知类频发的复杂场景，可支撑高分辨体制雷达对非合作目标的探测识别；同时，得益于良好的工程可实现性，该方法亦适用于机载平台等高实时性需求、硬件资源受限的应用场景。

Abstract

Radar target high-resolution range profile (HRRP) has significant application value in radar target recognition due to its ability to effectively characterize the geometric structure and electromagnetic scattering properties of targets. In recent years

deep learning methods have been widely applied to HRRP-based target recognition tasks owing to their powerful feature representation capability. However

existing deep learning-based radar target recognition methods using HRRP suffer from performance degradation in few-shot open-set recognition (FSOSR) scenarios due to limited training samples and inherent inability to discriminate unknown classes. To address this issue

this paper proposes a radar target HRRP few-shot open-set recognition method based on multi-hypersphere spatial topological constraints. First

a meta-learning framework is introduced in the training strategy. A large number of similar yet non-overlapping tasks are sampled from an auxiliary dataset with sufficient categories and abundant samples as training units to learn cross-task common knowledge

enabling the model to generalize rapidly on sample-scarce task datasets and alleviating few-shot overfitting. On this basis

a multi-hypersphere decision boundary modeling mechanism suitable for few-shot scenarios is designed. Each known class feature subspace is modeled by an individual hypersphere

forming a multi-hypersphere representation of the known-class feature space

which implicitly models the distribution of unknown classes. Meanwhile

adaptive margins are introduced between hyperspheres to maintain specific spatial topological relationships among known classes

thereby improving the refinement and robustness of the decision boundaries. Furthermore

a hard sample learning strategy is proposed. High-value samples are first selected through hard sample mining

after which a sample-wise weighting mechanism is employed to quantitatively characterize the hardness of samples. The model then performs targeted learning according to the assigned weights

enabling the extraction of finer discriminative features from informative hard samples and enhancing the model’s capability to distinguish fine-grained unknown classes. Experimental results demonstrate that

in the 5-shot and 10-shot scenarios

the proposed method achieves accuracy improvements of 6.17 percentage point and 2.94 percentage point

and AUROC improvements of 13.1 percentage point and 12.94 percentage point compared with the state-of-the-art methods

respectively

verifying the effectiveness and robustness of the proposed approach. In addition

the model is deployed on the Rockchip RK3588 embedded artificial intelligence (AI) chip

achieving an inference latency of 2.197 ms and a power consumption of 2.25 W

which demonstrates its engineering feasibility. The proposed method is suitable for complex environments where unknown classes frequently appear

and can support high-resolution radar systems in the detection and recognition of non-cooperative targets. Moreover

due to its favorable engineering feasibility

the method is also applicable to scenarios with stringent real-time requirements and limited hardware resources

such as airborne platforms.

关键词

Keywords

references

翟颖 , 陈渤 . 基于稳健变分自编码模型的雷达高分辨距离像目标识别算法 [J ] . 电子学报 , 2020 , 48 ( 6 ): 1149 - 1155 . DOI: 10.3969/j.issn.0372-2112.2020.06.015 http://dx.doi.org/10.3969/j.issn.0372-2112.2020.06.015

Zhai Ying , Chen Bo . Robust variational auto-encoder for radar HRRP target recognition [J ] . Acta Electronica Sinica , 2020 , 48 ( 6 ): 1149 - 1155 . (in Chinese) . DOI: 10.3969/j.issn.0372-2112.2020.06.015 http://dx.doi.org/10.3969/j.issn.0372-2112.2020.06.015

Wang J J , Liu Z , Ran L , et al . Feature extraction method for DCP HRRP-based radar target recognition via m -χ decomposition and sparsity-preserving discriminant correlation analysis [J ] . IEEE Sensors Journal , 2020 , 20 ( 8 ): 4321 - 4332 . DOI: 10.1109/jsen.2019.2962573 http://dx.doi.org/10.1109/jsen.2019.2962573

廖磊瑶 . 基于物理可解释自编码模型的雷达目标方位稳健识别算法 [J ] . 电子学报 , 2024 , 52 ( 11 ): 3847 - 3857 .

Liao Leiyao . High-range resolution radar target recognition based on deep interpretable probabilistic auto-encoder model [J ] . Acta Electronica Sinica , 2024 , 52 ( 11 ): 3847 - 3857 . (in Chinese)

Dong G G , Liu H W . A hierarchical receptive network oriented to target recognition in SAR images [J ] . Pattern Recognition , 2022 , 126 : 108558 . DOI: 10.1016/j.patcog.2022.108558 http://dx.doi.org/10.1016/j.patcog.2022.108558

杨威 , 李玮杰 , 刘永祥 , 等 . 基于测地线流式核的雷达目标高分辨距离像鲁棒识别方法 [J ] . 电子学报 , 2023 , 51 ( 3 ): 527 - 536 .

Yang Wei , Li Weijie , Liu Yongxiang , et al . Radar target recognition method for HRRP based on geodesic flow kernel [J ] . Acta Electronica Sinica , 2023 , 51 ( 3 ): 527 - 536 . (in Chinese)

Feng B , Chen B , Liu H W . Radar HRRP target recognition with deep networks [J ] . Pattern Recognition , 2017 , 61 : 379 - 393 . DOI: 10.1016/j.patcog.2016.08.012 http://dx.doi.org/10.1016/j.patcog.2016.08.012

Du L , Liu H W , Bao Z , et al . Radar HRRP target recognition based on higher order spectra [J ] . IEEE Transactions on Signal Processing , 2005 , 53 ( 7 ): 2359 - 2368 . DOI: 10.1109/tsp.2005.849161 http://dx.doi.org/10.1109/tsp.2005.849161

Du L , He H , Zhao L , et al . Noise robust radar HRRP target recognition based on scatterer matching algorithm [J ] . IEEE Sensors Journal , 2016 , 16 ( 6 ): 1743 - 1753 . DOI: 10.1109/jsen.2015.2501850 http://dx.doi.org/10.1109/jsen.2015.2501850

Li L , Liu Z , Li T . Radar high-resolution range profile feature extraction method based on multiple kernel projection subspace fusion [J ] . IET Radar , Sonar & Navigation, 2018 , 12 ( 4 ): 417 - 425 . DOI: 10.1049/iet-rsn.2017.0391 http://dx.doi.org/10.1049/iet-rsn.2017.0391

Lundén J , Koivunen V . Deep learning for HRRP-based target recognition in multistatic radar systems [C ] // 2016 IEEE Radar Conference . Piscataway : IEEE , 2016 : 1 - 6 . DOI: 10.1109/radar.2016.7485271 http://dx.doi.org/10.1109/radar.2016.7485271

Xu B , Chen B , Wan J W , et al . Target-aware recurrent attentional network for radar HRRP target recognition [J ] . Signal Processing , 2019 , 155 : 268 - 280 . DOI: 10.1016/j.sigpro.2018.09.041 http://dx.doi.org/10.1016/j.sigpro.2018.09.041

Du C , Chen B , Xu B , et al . Factorized discriminative conditional variational auto-encoder for radar HRRP target recognition [J ] . Signal Processing , 2019 , 158 : 176 - 189 . DOI: 10.1016/j.sigpro.2019.01.006 http://dx.doi.org/10.1016/j.sigpro.2019.01.006

郭泽坤 , 田隆 , 韩宁 , 等 . 采用CNN-SSD的雷达HRRP小样本目标识别方法 [J ] . 西安电子科技大学学报(自然科学版) , 2021 , 48 ( 2 ): 1 - 14 .

Guo Zekun , Tian Long , Han Ning , et al . Radar HRRP based few-shot target recognition with CNN-SSD [J ] . Journal of Xidian University (Natural Science) , 2021 , 48 ( 2 ): 1 - 14 . (in Chinese)

Huang Y , Wen Y , Shi L C , et al . Recognition-aware HRRP generation with generative adversarial network [J ] . IEEE Geoscience and Remote Sensing Letters , 2022 , 19 : 3503705 . DOI: 10.1109/lgrs.2021.3056192 http://dx.doi.org/10.1109/lgrs.2021.3056192

Liu Q , Zhang X Y , Liu Y X , et al . Few-shot HRRP target recognition based on gramian angular field and model-agnostic meta-learning [C ] // 2021 IEEE 6th International Conference on Signal and Image Processing . Piscataway : IEEE , 2021 : 6 - 10 . DOI: 10.1109/icsip52628.2021.9688842 http://dx.doi.org/10.1109/icsip52628.2021.9688842

Guo Z K , Liu Z , Xie R , et al . HRRP few-shot target recognition for full polarimetric radars via SCs optimal matching [J ] . IEEE Transactions on Aerospace and Electronic Systems , 2025 , 61 ( 2 ): 4526 - 4541 . DOI: 10.1109/taes.2024.3505117 http://dx.doi.org/10.1109/taes.2024.3505117

Chen W , Wang Y H , Song J , et al . Open set HRRP recognition based on convolutional neural network [J ] . The Journal of Engineering , 2019 , 2019( 21 ): 7701 - 7704 . DOI: 10.1049/joe.2019.0706 http://dx.doi.org/10.1049/joe.2019.0706

Xia Z H , Wang P H , Dong G G , et al . Surrounding prototype loss for radar HRRP open set target recognition [J ] . IEEE Geoscience and Remote Sensing Letters , 2022 , 19 : 4027305 . DOI: 10.1109/lgrs.2022.3211623 http://dx.doi.org/10.1109/lgrs.2022.3211623

Li W T , Li S , Tian B , et al . Regularization-based region learning for radar HRRP open set recognition [J ] . IEEE Transactions on Instrumentation and Measurement , 2025 , 74 : 2512615 . DOI: 10.1109/tim.2025.3546410 http://dx.doi.org/10.1109/tim.2025.3546410

Liu B , Kang H , Li H X , et al . Few-shot open-set recognition using meta-learning [C ] // 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2020 : 8795 - 8804 . DOI: 10.1109/cvpr42600.2020.00882 http://dx.doi.org/10.1109/cvpr42600.2020.00882

Pal D , Bose S , Banerjee B , et al . MORGAN: Meta-learning-based few-shot open-set recognition via generative adversarial network [C ] // 2023 IEEE/CVF Winter Conference on Applications of Computer Vision . Piscataway : IEEE , 2023 : 6284 - 6293 . DOI: 10.1109/wacv56688.2023.00623 http://dx.doi.org/10.1109/wacv56688.2023.00623

Wang H Y , Pang G S , Wang P , et al . Glocal energy-based learning for few-shot open-set recognition [C ] // 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2023 : 7507 - 7516 . DOI: 10.1109/cvpr52729.2023.00725 http://dx.doi.org/10.1109/cvpr52729.2023.00725

Jeong M , Choi S , Kim C . Few-shot open-set recognition by transformation consistency [C ] // 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2021 : 12561 - 12570 . DOI: 10.1109/cvpr46437.2021.01238 http://dx.doi.org/10.1109/cvpr46437.2021.01238

Ye H J , Hu H X , Zhan D C , et al . Few-shot learning via embedding adaptation with set-to-set functions [C ] // 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2020 : 8805 - 8814 . DOI: 10.1109/cvpr42600.2020.00883 http://dx.doi.org/10.1109/cvpr42600.2020.00883

Zaheer M , Kottur S , Ravanbakhsh S , et al . Deep sets [J ] . Advances in Neural Information Processing Systems , 2017 , 30 .

Wu C Y , Manmatha R , Smola A J , et al . Sampling matters in deep embedding learning [C ] // 2017 IEEE International Conference on Computer Vision . Piscataway : IEEE , 2017 : 2859 - 2867 . DOI: 10.1109/iccv.2017.309 http://dx.doi.org/10.1109/iccv.2017.309

Wang X , Han X T , Huang W L , et al . Multi-similarity loss with general pair weighting for deep metric learning [C ] // 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2019 : 5017 - 5025 . DOI: 10.1109/cvpr.2019.00516 http://dx.doi.org/10.1109/cvpr.2019.00516

Ruff L , Vandermeulen R , Goernitz N , et al . Deep one-class classification [C ] // International conference on machine learning . Cambridge : PMLR , 2018 : 4393 - 4402 .

Snell J , Swersky K , Zemel R . Prototypical networks for few-shot learning [J ] . Advances in Neural Information Processing Systems , 2017 , 30 .

Tax D M J , Duin R P W . Support vector data description [J ] . Machine Learning , 2004 , 54 ( 1 ): 45 - 66 . DOI: 10.1023/b:mach.0000008084.60811.49 http://dx.doi.org/10.1023/b:mach.0000008084.60811.49

唐金龙 . 基于图神经网络的雷达一维距离像识别 [D ] . 杭州 : 杭州电子科技大学 , 2023 .

Tang Jinlong . Radar One-Dimensional Range Profile Target Recognition Based on Graph Neural Network [D ] . Hangzhou : Hangzhou Dianzi University , 2023 . (in Chinese)

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于特征注意力融合元残差网络的小样本SAR目标识别

MSGPose：基于多语义图卷积与图引导状态空间模型的单目单人三维人体姿态估计

基于多视角置信度融合的对抗样本迁移性提升方法

面向医疗缺失数据的选择性学习

面向四天线反向散射标签的堆叠旋转Alamouti编码