A Multi-Objective Optimization Method in the Frequency Domain for SAR Image Adversarial Sample Generation

LIU Jie-yi; LI Ming-zhe; YANG Yao-ming; LI Hao; ZHOU Yu; DANG Ke-lin

doi:10.12263/DZXB.20250095

您当前的位置：

首页 >

文章列表页 >

A Multi-Objective Optimization Method in the Frequency Domain for SAR Image Adversarial Sample Generation

PAPERS | 更新时间：2025-10-16

- A Multi-Objective Optimization Method in the Frequency Domain for SAR Image Adversarial Sample Generation
- ACTA ELECTRONICA SINICA Vol. 53, Issue 6, Pages: 1958-1968(2025)
- 作者机构：
  
  西安电子科技大学电子工程学院，陕西西安 710071
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(62106185)
- DOI：10.12263/DZXB.20250095
  CLC： TN974;
- Received：06 February 2025，
  
  Revised：2025-05-29，
  
  Published：25 June 2025
- 稿件说明：
移动端阅览
刘洁怡, 李明哲, 杨曜铭, 等. 基于频域多目标优化的SAR图像对抗样本生成方法[J]. 电子学报, 2025, 53(06): 1958-1968.

LIU Jie-yi, LI Ming-zhe, YANG Yao-ming, et al. A Multi-Objective Optimization Method in the Frequency Domain for SAR Image Adversarial Sample Generation[J]. Acta Electronica Sinica, 2025, 53(06): 1958-1968.
刘洁怡, 李明哲, 杨曜铭, 等. 基于频域多目标优化的SAR图像对抗样本生成方法[J]. 电子学报, 2025, 53(06): 1958-1968. DOI：10.12263/DZXB.20250095

LIU Jie-yi, LI Ming-zhe, YANG Yao-ming, et al. A Multi-Objective Optimization Method in the Frequency Domain for SAR Image Adversarial Sample Generation[J]. Acta Electronica Sinica, 2025, 53(06): 1958-1968. DOI：10.12263/DZXB.20250095

摘要

基于深度学习的合成孔径雷达（Synthetic Aperture Radar，SAR）目标识别方法在军事侦察、灾害监测等领域应用广泛，然而深度神经网络易受到对抗攻击的威胁，导致模型决策的可靠性下降.现有黑盒对抗攻击方法在SAR图像对抗样本生成过程中面临参数设计维度高、易被察觉等问题.针对以上问题，提出一种基于频域多目标优化的对抗攻击方法，通过二维离散傅里叶变换将SAR图像从空间域映射至频域，降低扰动设计维度，进而在频域中修改单一频率分量，以生成图像域纹理状扰动.同时，结合基于超体积度量的多目标进化算法平衡对抗样本的攻击性能与视觉隐蔽性.实验结果表明，以T62类别为例，运用本文方法后，在VGG16、AConvNet和YOLO系列模型架构上，对抗样本分别实现了90.39%、71.43%、44.28%以上的置信度错误分类.同时，生成的对抗样本与原始图像的相似度均高于99%，为SAR图像的安全性与鲁棒性测试提供了有效的技术支持.

Abstract

Deep learning-based synthetic aperture radar (SAR) target recognition methods are widely used in military reconnaissance and disaster monitoring. However

deep neural networks (DNNs) are vulnerable to adversarial attacks

which compromise the reliability of model decisions. Existing black-box adversarial attack methods for SAR images face challenges such as high-dimensional parameter design and perceptible perturbations. To address these issues

a frequency-domain multi-objective optimization-based adversarial attack method is proposed. By transforming SAR images from the spatial domain to the frequency domain via 2D Discrete Fourier Transform

the method reduces perturbation design complexity and modifies a single frequency component to generate texture-like perturbations in the spatial domain. A hypervolume metric-guided multi-objective evolutionary algorithm is integrated to balance attack performance and visual stealthiness. Experimental results demonstrate that

for the T62 category

the adversarial samples generated by our method achieve misclassification confidence rates of more than 90.39%

71.43%

44.28% on VGG16

AConvNet

and YOLO series models

respectively. Additionally

the similarity between adversarial and original images exceeds 99% across all cases

providing effective technical support for security and robustness evaluation of SAR imaging systems.

关键词

Keywords

references

XIA W J , LIU Z , LI Y . SAR-PeGA: A generation method of adversarial examples for SAR image target recognition network [J ] . IEEE Transactions on Aerospace and Electronic Systems , 2022 , 59 ( 2 ): 1910 - 1920 .

化青龙 , 张云 , 任航 , 等 . 基于最小熵准则与生成对抗网络的SAR三维转动舰船目标重聚焦方法 [J ] . 电子学报 , 2024 , 52 ( 8 ): 2900 - 2912 .

HUA Q L , ZHANG Y , REN H , et al . Refocusing for three-dimensional rotating ship targets in SAR images based on minimum entropy criteria and generative adversarial network [J ] . Acta Electronica Sinica , 2024 , 52 ( 8 ): 2900 - 2912 . (in Chinese)

SHI J . SAR target recognition method of MSTAR data set based on multi-feature fusion [C ] // 2022 International Conference on Big Data, Information and Computer Network . Piscataway : IEEE , 2022 : 626 - 632 .

WU Y , LIU J M , GONG M G , et al . Joint Semantic Segmentation using representations of LiDAR point clouds and camera images [J ] . Information Fusion , 2024 , 108 : 102370 .

化青龙 , 魏晨曦 , 张云 , 等 . 基于自监督复数域深度学习网络的SAR有源压制干扰抑制方法 [J ] . 电子学报 , 2023 , 51 ( 4 ): 965 - 974 .

HUA Q L , WEI C X , ZHANG Y , et al . Active jamming suppression for SAR images based on self-supervised complex-valued deep learning [J ] . Acta Electronica Sinica , 2023 , 51 ( 4 ): 965 - 974 . (in Chinese)

XIE C H , ZHANG Z S , ZHOU Y Y , et al . Improving transferability of adversarial examples with input diversity [C ] // 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2019 : 2725 - 2734 .

MOOSAVI-DEZFOOLI S M , FAWZI A , FROSSARD P . DeepFool: A simple and accurate method to fool deep neural networks [C ] // 2016 IEEE Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2016 : 2574 - 2582 .

CARLINI N , WAGNER D . Towards evaluating the robustness of neural networks [C ] // 2017 IEEE Symposium on Security and Privacy . Piscataway : IEEE , 2017 : 39 - 57 .

LI H F , HUANG H K , CHEN L , et al . Adversarial examples for CNN-based SAR image classification: An experience study [J ] . IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2020 , 14 : 1333 - 1347 .

DU C , HUO C Y , ZHANG L , et al . Fast CW: A fast adversarial attack algorithm to fool SAR target recognition with deep convolutional neural networks [J ] . IEEE Geoscience and Remote Sensing Letters , 2021 , 19 : 4010005 .

ANDRIUSHCHENKO M , CROCE F , FLAMMARION N , et al . Square Attack: A Query-Efficient Black-Box Adversarial Attack via Random Search [M ] // Computer Vision-ECCV 2020 . Cham : Springer International Publishing , 2020 : 484 - 501 .

GUO C , GARDNER J , YOU Y Y , et al . Simple black-box adversarial attacks [C ] // International Conference on Machine Learning . Long Beach : PMLR , 2019 : 2484 - 2493 .

DU C , ZHANG L . Adversarial attack for SAR target recognition based on UNet-generative adversarial network [J ] . Remote Sensing , 2021 , 13 ( 21 ): 4358 .

CUI J H , DUAN J L , LUO B Y , et al . SAR-AE-SFP: SAR imagery adversarial example in real physics domain with target scattering feature parameters [EB /OL ] . ( 2024-03-02 )[ 2025-05-29 ] . https://doi.org/10.48550/arXiv.2403.01210 https://doi.org/10.48550/arXiv.2403.01210 .

ZHANG L , JIANG T P , GAO S Y , et al . Generating adversarial examples on sar images by optimizing flow field directly in frequency domain [C ] // IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium . Piscataway : IEEE , 2022 : 2979 - 2982 .

HUANG X C , LU Z Z , PENG B . Frequency and spatial adversarial attack to SAR target recognition [C ] // 2024 7th International Conference on Electronics Technology . Piscataway : IEEE , 2024 : 906 - 910 .

TESSARI G , FLORIS M , PASQUALI P . Phase and amplitude analyses of SAR data for landslide detection and monitoring in non-urban areas located in the North-Eastern Italian pre-Alps [J ] . Environmental Earth Sciences , 2017 , 76 ( 2 ): 85 .

BEUME N , NAUJOKS B , EMMERICH M . SMS-EMOA: Multiobjective selection based on dominated hypervolume [J ] . European Journal of Operational Research , 2007 , 181 ( 3 ): 1653 - 1669 .

LIN T W , MA Z Q , LI F , et al . Drafting and revision: Laplacian pyramid network for fast high-quality artistic style transfer [C ] // 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2021 : 5137 - 5146 .

TAO M L , SU J , HUANG Y , et al . Mitigation of radio frequency interference in synthetic aperture radar data: Current status and future trends [J ] . Remote Sensing , 2019 , 11 ( 20 ): 2438 .

JIANG L M , DAI B , WU W , et al . Focal frequency loss for image reconstruction and synthesis [C ] // 2021 IEEE/CVF International Conference on Computer Vision . Piscataway : IEEE , 2021 : 13899 - 13909 .

陈思伟 , 周鹏 . SAR图像对抗攻击的进展与展望 [J ] . 信息对抗技术 , 2023 , 2 ( Z1 ): 171 - 188 .

CHEN S W , ZHOU P . Progress and prospect of SAR image adversarial attack [J ] . Information Countermeasure Technology , 2023 , 2 ( Z1 ): 171 - 188 . (in Chinese)

BEUME N , FONSECA C M , LOPEZ-IBANEZ M , et al . On the complexity of computing the hypervolume indicator [J ] . IEEE Transactions on Evolutionary Computation , 2009 , 13 ( 5 ): 1075 - 1082 .

章坚武 , 能豪 , 李杰 , 等 . 基于掩模提取的SAR图像对抗样本生成方法 [J ] . 电信科学 , 2024 , 40 ( 3 ): 64 - 74 .

ZHANG J W , NAI H , LI J , et al . Adversarial example generation method for SAR images based on mask extraction [J ] . Telecommunications Science , 2024 , 40 ( 3 ): 64 - 74 . (in Chinese)

SIMONYAN K , ZISSERMAN A . Very deep convolutional networks for large-scale image recognition [EB/OL ] . ( 2015-04-10 )[ 2025-05-29 ] . https://doi.org/10.48550/arXiv.1409.1556 https://doi.org/10.48550/arXiv.1409.1556 .

LIU Z , MAO H Z , WU C Y , et al . A ConvNet for the 2020s [C ] // 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2022 : 11966 - 11976 .

ZHANG J , HUO Y B , YANG J L , et al . Automatic counting of retinal ganglion cells in the entire mouse retina based on improved YOLOv5 [J ] . Zoological Research , 2022 , 43 ( 5 ): 738 - 749 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A Multi-Layer Feature Importance Attack Method Based on Iterative Accumulated Gradients

Related Author

WU Ji

SHAO Wen-ze

GE Qi

SUN Yu-bao

Related Institution

School of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications

Engineering Research Center for Digital Forensics Ministry of Education, Nanjing University of Information Science and Technology

⁰