摘要
针对人脸图像数字信息易遭受非法篡改、编造的安全性问题,本文提出了一种人脸伪造主动防御的方法。首先运用基于肤色分割的人脸检测技术识别目标图片中的人脸,然后使用DCT域的数字水印方法,将检测识别的人脸作为水印嵌入到目标图像中,以达到主动防御的目的,最后提取图片中嵌入的水印,运用相关系数和峰值信噪比等指标进行评价。实验结果表明,与规定好的阈值相比,抵御攻击的成功率达98%,该方法可主动抵御一些较为常见的攻击,满足设计的基本要求。
In response to the security issue of facial image digital information being susceptible to illegal tampering and fabrication,this paper proposes an active defense method for facial forgery.Firstly,facial detection technology based on skin color segmentation is used to identify the faces in the target image.Then,a digital watermarking method in the DCT domain is used to embed the detected and recognized faces as watermarks into the target image for active defense.Finally,the watermarks embedded in the image are extracted and evaluated using indicators such as correlation coefficient and peak signal-to-noise ratio.The experimental results show that compared to the specified threshold,the success rate of resisting attacks reaches 98%.This method can actively resist some common attacks and meet the basic requirements of the design.
作者
王清波
李振
王一诺
WANG Qingbo;LI Zhen;WANG Yinuo(Shandong Institute of Product Quality Inspection,Jinan,Shandong 250102,China;School of Information and Electrical Engineering,Shandong Jianzhu University,Jinan,Shandong 250101,China)
出处
《自动化应用》
2023年第20期172-175,178,共5页
Automation Application
关键词
主动防御
人脸检测
DCT域变换
数字水印
active defense
face recognition
DCT domain transform
digital watermarking
作者简介
王清波,男,1984年生,硕士,工程师,从事计算机和木材检验检测相关技术研究;通讯作者:王一诺,男,2000年生,硕士研究生,研究方向为智能环境与网络化控制。
