A Survey of Image Forgery Detection
A Survey of Image Forgery Detection
复制标题
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
H. Farid
中科院分区：
文献类型：
--
作者：
H. Farid
当前位置毫无疑问，我们生活在一个接触到大量视觉意象的时代。虽然我们在历史上可能对这些图像的完整性有信心，但今天的数字技术已经开始侵蚀这种信任。从小报杂志到时尚行业、主流媒体、科学期刊、政治竞选、法庭，以及我们电子邮件收件箱里的照片骗局，伪造照片的出现频率越来越高，手法也越来越复杂。在过去的五年中，数字取证领域的出现有助于恢复对数字图像的信任。在这里，我回顾了这个令人兴奋的新领域的最新进展。数字水印已被提议作为一种可以对图像进行身份验证的手段（例如，参见[21,5]进行一般调查）。这种方法的缺点是必须在记录时插入水印，这将限制这种方法用于专门配备的数码相机。与这些方法相比，被动图像取证技术在没有任何水印或签名的情况下运行。这些技术的工作原理是基于这样一个假设：尽管数字伪造可能不会留下被篡改的视觉线索，但它们可能会改变图像的基本统计数据。图像取证工具集大致可分为五类：(1)基于像素的技术检测在像素级引入的统计异常；(2)基于格式的技术利用了特定有损压缩方案引入的统计相关性；(3)基于相机的技术利用相机镜头、传感器或片上后处理引入的伪影；(4)基于物理的技术明确地模拟和检测物理对象、光线和相机之间的三维相互作用中的异常；(5)基于几何的技术测量世界中的物体及其相对于相机的位置。我在这些类别中选择了几个有代表性的取证工具来进行审查。在这样做的过程中，我无疑遗漏了一些有价值的论文。然而，我的希望是，这项调查为图像伪造检测的新兴领域提供了一个有代表性的样本。
: We are undoubtedly living in an age where we are exposed to a remarkable array of visual imagery. While we may have historically had confidence in the integrity of this imagery, today’s digital technology has begun to erode this trust. From the tabloid magazines to the fashion industry, main-stream media outlets, scientific journals, political campaigns, courtrooms, and the photo hoaxes that land in our email in-boxes, doctored photographs are appearing with a growing frequency and sophistication. Over the past five years, the field of digital forensics has emerged to help return some trust to digital images. Here I review the state of the art in this new and exciting field. Digital watermarking has been proposed as a means by which an image can be authenticated (see, for example, [21, 5] for general surveys). The drawback of this approach is that a watermark must be inserted at the time of recording, which would limit this approach to specially equipped digital cameras. In contrast to these approaches, passive techniques for image forensics operate in the absence of any watermark or signature. These techniques work on the assumption that although digital forgeries may leave no visual clues of having been tampered with, they may alter the underlying statistics of an image. The set of image forensic tools can be roughly categorized into five categories: (1) pixel-based techniques detect statistical anomalies introduced at the pixel level; (2) format-based techniques leverage the statistical correlations introduced by a specific lossy compression scheme; (3) camera-based techniques exploit artifacts introduced by the camera lens, sensor or on-chip post-processing; (4) physically-based techniques explicitly model and detect anomalies in the three dimensional interaction between physical objects, light, and the camera; and (5) geometric-based techniques make measurements of objects in the world and their positions relative to the camera. I have selected several representative forensic tools within each of these categories to review. In so doing, I have undoubtedly omitted some worthy papers. My hope, however, is that this survey offers a representative sampling of the emerging field of image forgery detection.