面向真实性鉴别的数字图像盲取证技术综述  

A Survey of Blind Digital Image Forensics Technology for Authenticity Detection

作　　者：吴琼[1] 李国辉[1] 涂丹[1] 孙韶杰[1] 

机构地区：[1]国防科技大学信息系统与管理学院,长沙410073

出　　处：《自动化学报》2008年第12期1458-1466,共9页Acta Automatica Sinica

摘　　要：数字图像盲取证技术作为一种不依赖任何预签名提取或预嵌入信息来鉴别图像真伪和来源的技术,正逐步成为多媒体安全领域新的研究热点,且有着广泛的应用前景.首先简要描述了图像盲取证技术要解决的问题和任务.根据图像鉴别使用的取证特征,将用于真实性鉴别的图像盲取证技术划分为三类:基于图像伪造过程遗留痕迹的盲取证技术、基于成像设备一致性的盲取证技术和基于自然图像统计特性的盲取证技术,然后分别阐述了这三类取证技术的基本特征和典型方法,对不同算法进行了性能比较和总结.最后综合近年来国内外学者在面向真实性鉴别的图像盲取证技术方面的主要研究成果,探讨了图像盲取证技术存在的问题及未来研究方向.Blind digital image forensics, a technology for detecting image authenticity and source without relying on any pre-extraction or pre-embedded information, is emerging as a new hotspot with broad prospect in the multimedia security field. First, the main problems and tasks of image forensics are briefly reviewed. Then, according to the forensics characteristics used by image authentication, the techniques of blind image forensics for authenticity detection are divided into three categories： techniques based on the traces left by the process of image forgery, techniques based on the consistency of imaging equipment, and techniques based on the statistical characteristics of natural images. The basic characteristics, typical methods, as well as performance comparison and analysis of various algorithms are summarized in detail for each category. So the latest related works are analyzed, and discussion on unresolved problems and future directions is presented.

关 键 词：真实性鉴别 图像盲取证 篡改检测 多媒体安全 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]