信息安全研究2025，Vol.11Issue(10)：941-949,9.DOI:10.12379/j.issn.2096-1057.2025.10.09
不可感知的人脸属性编辑伪造主动防御方法
Imperceptible Proactive Defense Method Against Face Attribute Editing
摘要
Abstract
Although the face attribute editing forgery active defense method based on generative adversarial network(GAN)generates adversarial perturbations faster than the gradient attack-based methods,existing methods still fail in balancing the proactive defense effect with the imperceptibility of generated perturbations.Therefore,this paper proposed a highly imperceptible proactive defense method against face attribute editing based on GAN.To enhance the imper-ceptibility of the perturbations,the method designed a high-frequency information compensation mechanism to enable the generator to generate more high-frequency perturbations that are less sensitive to the human eye.To improve the proactive defense performance of generated pertur-bations,the proposed method also designed a multi-level dense connection mechanism for reducing semantic loss during the encoding process.Meanwhile,the method introduced face saliency adversarial loss in training stage to enable perturbations to disrupt face forgery areas better.The experiments were conducted in both single-model and cross-model defense scenarios.The results indicate that compared to existing methods,the proposed method generates more imperceptible adversarial perturbations and obtains high success rates for defending against target models.关键词
深度伪造/对抗样本/主动防御/生成对抗网络/不可感知性Key words
deepfake/adversarial example/proactive defense/generative adversarial network/imperceptibility分类
信息技术与安全科学引用本文复制引用
陈北京,冯逸凡,范春年..不可感知的人脸属性编辑伪造主动防御方法[J].信息安全研究,2025,11(10):941-949,9.基金项目
国家自然科学基金项目(62072251) （62072251）
