数字图像操作取证综述  

A Survey of Digital Image Operation Forensics

作　　者：谢皓 张健[2] 倪江群[2] XIE Hao;ZHANG Jian;NI Jiangqun(School of Electronics and Information Technology,Sun Yat-sen University,Guangzhou,Guangdong 510006,China;School of Computer Science and Engineering,Sun Yat-sen University,Guangzhou,Guangdong 510006,China)

机构地区：[1]中山大学电子与信息工程学院,广东广州510006 [2]中山大学计算机学院,广东广州510006

出　　处：《信号处理》2021年第12期2323-2337,共15页Journal of Signal Processing

基　　金：国家自然科学基金(U1936212,U1736215,61772573)。

摘　　要：如今,数字图像编辑技术和软件已经被广泛应用于各行各业,但是图像编辑技术被滥用的案例也层出不穷,造成严重的后果与不良的社会影响。为了证明和维护数字图像的真实性,防止图像编辑技术被滥用,数字图像取证技术已成为与社会生活息息相关的重要技术和多媒体信息安全领域的重要研究方向。本文针对数字图像取证中图像操作检测任务,归纳图像操作检测的基本原理,并介绍在单一操作检测、通用操作检测、操作链检测等多个取证场景下,操作痕迹分析、传统手工取证特征以及深度学习检测器等的最新研究进展,最后对当前图像操作取证研究和挑战进行了总结,并对今后可能的研究方向做了展望。Nowadays, the digital image editing techniques and tools are widely used in various fields in our society, but the increasing cases of image malicious tampering also show the serious consequences of the easy access to image editing. In order to verify and maintain the authenticity of digital images and prevent the abuse of image editing techniques, digital image forensics are catching more and more attention and becoming a hot topic in multi-media information security. Based on the studies on image operation detection in digital image forensics, this paper briefly reviews the basic ideas of image operation detection at first. Then we introduce the latest researches of the operation trace analysis, traditional hand-crafted features, and deep learning detectors under various forensic scenarios, including single operation detection, general operation detection and operation chain detection. Finally, the conclusions and discussions are made on the existing challenges and future developments for image operation detection.

关 键 词：图像操作检测 数字图像取证 多媒体信息安全 深度学习 操作链检测 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]