基于ICA运动目标检测的自嵌入视频认证水印  

Self-embedded watermarking scheme for video authentication based on ICA motion detection

作　　者：张婷[1] 吕皖丽[1] 罗斌[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039

出　　处：《计算机工程与设计》2011年第4期1370-1373,1390,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(60772122);安徽高等学校省级自然科学研究重点基金项目(KJ2009A145)

摘　　要：为了对视频中关键内容是否经受篡改实施有效验证和保护,提出了一种基于运动目标检测的自嵌入视频认证水印方案。由于运动目标集中了视频的关键信息,采用独立分量分析技术从连续的视频帧中分离出运动分量信息生成水印,并用奇异值分解方法对水印作预处理,采用DCT方法自嵌入到视频帧中去。实验结果表明,该水印方案在保证视频质量的前提下,对于常见攻击处理和局部篡改具有很好的鲁棒性,可以有效地验证视频关键信息是否遭到篡改,并且能够恢复出被篡改的关键内容,有效地保护视频完整性。To find out whether the important content of video had been partly tempered and protect the crucial information,a novel self-embedded video watermarking scheme is proposed based on motion detection.The motion objects attract humans’ attention and present as the major content of video,so independent component analysis is used to extract motion information as our watermark from successive frames of original video.Then the watermark is embedded in the DCT domain which is pre-processed by SVD into the video frames.In the premise of ensuring the video quality,the simulations show that the proposed scheme has a good performance to resist common signal processing attacks and local changes,so the results demonstrate that the algorithm could check whether the video had been atta-cked,and recover the major content very well when the watermarked video is partly tampered to protect the completeness of the video.

关 键 词：视频认证 独立分量分析 奇异值分解 内容保护 自嵌入水印 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]