Adversarial example generation method for SAR images based on mask extraction
There are many ways to generate adversarial samples for synthetic aperture radar (SAR) images at present, but some problems such as large amount of perturbation of adversarial samples, unstable training, and unguaranteed quality of adversarial samples still exist. To solve the above problems, a SAR...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Beijing Xintong Media Co., Ltd
2024-03-01
|
| Series: | Dianxin kexue |
| Subjects: | |
| Online Access: | http://www.telecomsci.com/zh/article/doi/10.11959/j.issn.1000-0801.2024081/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850091170063974400 |
|---|---|
| author | ZHANG Jianwu NAI Hao LI Jie QIAN Jianhua FANG Yinfeng |
| author_facet | ZHANG Jianwu NAI Hao LI Jie QIAN Jianhua FANG Yinfeng |
| author_sort | ZHANG Jianwu |
| collection | DOAJ |
| description | There are many ways to generate adversarial samples for synthetic aperture radar (SAR) images at present, but some problems such as large amount of perturbation of adversarial samples, unstable training, and unguaranteed quality of adversarial samples still exist. To solve the above problems, a SAR image adversarial sample generation model was proposed. The model was based on the AdvGAN model architecture. Firstly, according to the characteristics of the SAR images, an adaptive threshold segmentation method based on the enhanced Lee filter OTSU was designed. The mask extraction module composed of equal modules, this method produced a smaller amount of disturbance, and the structural similarity (SSIM) with the original sample reached that more than 0.997. Secondly, the improved relativistic average GAN (RaGAN) loss was introduced into AdvGAN, and the relative mean discriminator was used to make the discriminator rely on both real data and generated data during training, which improved the stability of training and the attack effect. Experiments were compared with related methods on the MSTAR dataset. Experiments show that the attack success rate of SAR image adversarial samples generated by this method is increased by 10%～15% than that of traditional methods when attacking defense models. |
| format | Article |
| id | doaj-art-378fb8c7478544fab9d1152a9d2a8732 |
| institution | DOAJ |
| issn | 1000-0801 |
| language | zho |
| publishDate | 2024-03-01 |
| publisher | Beijing Xintong Media Co., Ltd |
| record_format | Article |
| series | Dianxin kexue |
| spelling | doaj-art-378fb8c7478544fab9d1152a9d2a87322025-08-20T02:42:26ZzhoBeijing Xintong Media Co., LtdDianxin kexue1000-08012024-03-0140647455075429Adversarial example generation method for SAR images based on mask extractionZHANG JianwuNAI HaoLI JieQIAN JianhuaFANG YinfengThere are many ways to generate adversarial samples for synthetic aperture radar (SAR) images at present, but some problems such as large amount of perturbation of adversarial samples, unstable training, and unguaranteed quality of adversarial samples still exist. To solve the above problems, a SAR image adversarial sample generation model was proposed. The model was based on the AdvGAN model architecture. Firstly, according to the characteristics of the SAR images, an adaptive threshold segmentation method based on the enhanced Lee filter OTSU was designed. The mask extraction module composed of equal modules, this method produced a smaller amount of disturbance, and the structural similarity (SSIM) with the original sample reached that more than 0.997. Secondly, the improved relativistic average GAN (RaGAN) loss was introduced into AdvGAN, and the relative mean discriminator was used to make the discriminator rely on both real data and generated data during training, which improved the stability of training and the attack effect. Experiments were compared with related methods on the MSTAR dataset. Experiments show that the attack success rate of SAR image adversarial samples generated by this method is increased by 10%～15% than that of traditional methods when attacking defense models.http://www.telecomsci.com/zh/article/doi/10.11959/j.issn.1000-0801.2024081/adversarial samplegenerative adversarial networksynthetic aperture radarsemi-white box attackmask extraction |
| spellingShingle | ZHANG Jianwu NAI Hao LI Jie QIAN Jianhua FANG Yinfeng Adversarial example generation method for SAR images based on mask extraction Dianxin kexue adversarial sample generative adversarial network synthetic aperture radar semi-white box attack mask extraction |
| title | Adversarial example generation method for SAR images based on mask extraction |
| title_full | Adversarial example generation method for SAR images based on mask extraction |
| title_fullStr | Adversarial example generation method for SAR images based on mask extraction |
| title_full_unstemmed | Adversarial example generation method for SAR images based on mask extraction |
| title_short | Adversarial example generation method for SAR images based on mask extraction |
| title_sort | adversarial example generation method for sar images based on mask extraction |
| topic | adversarial sample generative adversarial network synthetic aperture radar semi-white box attack mask extraction |
| url | http://www.telecomsci.com/zh/article/doi/10.11959/j.issn.1000-0801.2024081/ |
| work_keys_str_mv | AT zhangjianwu adversarialexamplegenerationmethodforsarimagesbasedonmaskextraction AT naihao adversarialexamplegenerationmethodforsarimagesbasedonmaskextraction AT lijie adversarialexamplegenerationmethodforsarimagesbasedonmaskextraction AT qianjianhua adversarialexamplegenerationmethodforsarimagesbasedonmaskextraction AT fangyinfeng adversarialexamplegenerationmethodforsarimagesbasedonmaskextraction |