摘要
Deepfake(深度伪造)技术可以生成人类肉眼无法辨别的虚假图像、视频,Deepfake与社交网络结合会给个人和社会带来极大的信息安全隐患.因此,能够自动检测和评估数字视觉媒体完整性的技术方案就显得尤为重要.从深度学习检测算法、数字来源取证和生命日志记录3个方向,回顾了Deepfake检测技术的发展现状,分析了不同反制方案面临的风险与挑战.研究表明,目前没有一个完美的解决方案能够有效应对Deepfake技术的现实威胁.最后,立足我国基本国情,对我国Deepfake检测技术的发展提出了建议.
Deepfake(deep forgery)technology can generate false images and videos that cannot be distinguished by human eyes.The combination of deep forgery and social networks will bring great information security risks to individuals and society.Therefore,it is essential to automatically detect and evaluate the integrity of digital visual media.This paper reviews the development of deepfake detection technology from three directions:deep learning detection algorithms,digital source forensics,and life log records,and analyzes the risks and challenges faced by different countermeasures.Research shows that there is currently no perfect solution to effectively deal with the real threat of deepfake technology.Finally,based on China’s basic national conditions,the development of China’s deepfake detection technology is proposed.
作者
白国柱
王蓓蓓
Bai Guozhu;Wang Beibei(National Computer Network and Information Security Management Center,Shaanxi Branch,Xi'an 710075;Shaanxi Information and Communication Industry Association,Xi'an 710075)
出处
《信息安全研究》
2020年第9期818-822,共5页
Journal of Information Security Research
关键词
深度伪造
检测技术
研究现状
挑战
启示
Deepfake
detection technology
research status
challenge
enlightenment
