一种利用变分辨率敏感度不一致性的对抗性人脸检测方法  

An Adversarial Face Detection Method Using Inconsistency Sensitivity Inconsistent of Variable Resolution

作　　者：刘宏基 黄瑞阳[1] 汪浣沙 LIU Hongji;HUANG Ruiyang;WANG Huansha(National Digital Switching System Engineering Technological R&D Center,Zhengzhou 450002,China)

机构地区：[1]国家数字交换系统工程技术研究中心,河南郑州450002

出　　处：《信息工程大学学报》2025年第2期175-181,共7页Journal of Information Engineering University

基　　金：国家重点研发计划(2022YFB4500900);江苏省科学技术厅项目(ZL042401);军队后勤科研项目(BHQ090003000X03)。

摘　　要：针对现有对抗性人脸检测方法在不同数据集间迁移性不足的问题,提出一种基于图像变分辨率敏感度不一致性的对抗性人脸检测方法。通过分析超分辨率、去噪等图像变换过程中的敏感度变化,利用余弦相似度量化变换前后人脸图像特征的差异,检测对抗性人脸攻击。在人脸数据库(LFW)、CelebA-HQ等数据集上的实验证明,该方法在6种对抗攻击条件下均达到90%以上的检测准确率。To address the limited transferability of existing adversarial face detection methods across datasets,a novel method based on the inconsistency of sensitivity to variable image resolution is proposed.The sensitivity changes during image transformations,such as super-resolution and denoising,are analyzed,and cosine similarity is employed to quantify the differences in facial features before and after transformation,enabling the detection of adversarial face attacks.Experiments on datasets such as Labeled Faces in the Wild(LFW)and CelebA-HQ have demonstrated that a detection accuracy rate of over 90%under six types of adversarial attacks is achieved by the proposed method.

关 键 词：深度学习安全 对抗样本检测 图像变换 人脸识别 超分辨率 图像去噪 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]