数字图像的盲取证技术由于不依赖任何预嵌入的信息来鉴别图像真实性和完整性的优势,正逐步成为数字媒体安全领域新的研究热点。由于JPEG图像是目前最流行的图像格式,并且块效应是JPEG图像与生俱来的本质特征,因此如何更加有效地利用块效应特征对JPEG图像的真伪进行盲取证研究具有非常重要的现实意义和应用价值。首先对目前国内外利用JPEG图像编码特性的盲取证方法进行归类分析;然后重点针对利用块效应特征的JPEG图像盲取证技术展开讨论,详细介绍并总结了基于块效应测度和基于块效应网格提取的两类盲取证算法的核心思想和局限性;最后提出了存在的问题及未来的研究方向。

Blind digital image forensics, a technology for detecting image authenticity and integrity without relying on any pre-embedded information, is emerging as a new hotspot in the digital media security filed. Since JPEG is the most popular image format and blocking artifacts are the inherent characteristics of JPEG images, using blocking artifacts more effectively for the authentication of JPEG image forgeries has very important practical significance and application value. First, the state-of-the-art algorithms of blind forensics based on JPEG image coding characteristics are classified and analyzed. Then we focus on blind forensic algorithms of tempered JPEG images based on blocking artifacts. Two types are introduced and summarized in detail: one based on blocking artifacts measurement and the other based on blocking artifacts grid extraction. Finally, we discuss unresolved problems and future research directions.