可视媒体取证新技术研究

项目名称： 可视媒体取证新技术研究

项目编号： No.61272191

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 骆伟祺

作者单位： 中山大学

项目金额： 82万元

中文摘要： 随着各种媒体处理方法及其软件的发展, 普通用户都可很简单地实现对媒体篡改且不留下视觉上痕迹。如今，眼见并不为实，媒体取证分析正面临严峻挑战。本项目将针对可视媒体（图像与视频）取证中的若干新技术进行深入探讨，其主要创新点包括：1.目前大部分取证方法仅利用了图像的低层视觉信息。项目拟通过人的简单交互，充分结合图像中高层视觉信息，以提高现有算法的检测性能；2.无缝篡改在计算机视觉、图形学中已研究多年，一些较成熟算法亦已添加到Photoshop新版本中，但相应的取证技术却尚未见报道。项目将首次分析图像/视频的无缝篡改；3.现有重压缩检测技术无法判别经不同源编码器转换后的图像/视频，项目拟研究不同源编码器的转换检测及参数估计；4.基于模式噪声与颜色插值特征的相机源识别算法有许多，但从反取证角度出发的研究报告且极少。项目拟提出一种图像内容自适应的模式噪声、颜色插值重嵌入策略，以混淆现有相机源识别算法。

中文关键词： 图像篡改；被动取证；反取证；；

英文摘要： Recent development in digital multimedia processing methods and related editing softwaretools, such as Adobe Photoshop and Premiere,has made it increasingly easy for ordinary users to tamper the contents of multimedia data without leaving obvious visual traces. Nowaday, seeing is no longer believing. And it will get worse as counterfeiting techniques get more and more sophisticated. Authentication of multimedia data faces great challenges.In this project, we focus on several new techniques in the visual media (including digital image and video) forensics, and the innovation of the project is list as follows: 1. Most existing works just employ the early vision information in an image for the purpose of multimedia forensics. Based on the reported results, it is found that their performances are far from satisfactory, and they usually are sensitive to some post-operations. In this project, we try to propose an interactive forensic scheme to make full use of the mid-level and high-level vision information for further improving the related works on image splicing and light consistency detection. 2. Seamless tampering (including object splicing and removal) is well studied in the research areas of computer vision and computer graphics. Some technologies ready and mature for application have already been included in

英文关键词： Image Tampering；Passive Forensics；Anti-forensics；；