基于内容的视频取证及拷贝检测关键技术的研究

With the adventure of easy-to-use video editing softwares,problems about video copyright and authentication of video integrity surged. Content-based video forensics and video copy detection, as newly developing technologies of digital media security, are attracting more and more attention in academic research. However, there is little research reported in these fields. Thus, there are many chanllengies and chances. While video can be seen as an orderly sequence of images, video processing has its own sepcial requirements and inherent characteristics compared with image processing. Based on deep analysis on sepcial requirements and inherent characteristics of video processing, video encoding and decoding process, video capturing and conversion process and video tampering methods, we research on new methods to improve the abilities of content-based video forensics and video copy detection. Specifically, it includes four aspects: 1) research on statistical characteristics and forensics algorithms for double H.264 compressed video; 2) research on interframe tampering evidence and forencics schemes for H.264 videos; 3) research on features for content-based video copy detection with good robustness, discriminability and compactness. 4) research on highly efficient search structure for content-based video copy detection. By researching on these issues, we try to develop new theories and algorithms of both video forencics and video copy detection and provide supporting key technologies.

由于简易的视频编辑软件层出不穷，视频的版权、视频的真实性等问题相继出现。基于内容的视频取证及视频拷贝检测作为一种数字媒体安全的新技术，已经成为备受瞩目的研究领域。但目前这两方面的研究工作都相对较少，充满了挑战和机遇。虽然视频可以看作是有序的图像序列，但视频处理有其特殊的要求和固有的特征。本项目拟在对基于内容的视频处理的特殊要求、固有特征、视频编解码过程、视频采集转换及视频篡改方式等进行深入分析的基础上，研究新的方法提高基于内容的视频取证及视频拷贝检测的能力，具体包括四个方面：1）H.264重压缩视频统计特征与取证算法的研究；2）H.264压缩视频帧间篡改的特征与取证算法的研究；3）具有良好鲁棒性、区分性、和紧凑性的用于视频拷贝检测的特征研究；4）高效的基于内容的视频拷贝检测搜索结构的研究。通过对上述问题的研究，发展和完善基于内容的视频取证、视频拷贝检测的理论和方法，并提供关键技术方面的支持

由于简易的视频编辑软件层出不穷，视频的版权、视频的真实性等问题相继出现。基于内容的视频取证作为一种数字媒体安全的新技术，已经成为备受瞩目的研究领域。但目前这方面的研究工作相对较少，充满了挑战和机遇。虽然视频可以看作是有序的图像序列，但视频处理有其特殊的要求和固有的特征。本项目针对H.264/AVC压缩的特殊性，1）提出了H.264/AVC 重压缩视频统计特征与取证算法；2）提出了H.264/AVC 压缩视频帧间篡改的特征与取证算法；3）研究H.264/AVC压缩视频PSNR的盲估计问题；4）继续研究视频的基础组成图像方面的信息安全算法。通过对上述问题的研究，发表论文4篇，申请专利2项, 通知授权发明专利1项。通过对上述问题的研究， 发展和完善基于内容的视频取证及图像信息安全的理论和方法，并提供关键技术方面的支持