【作者】 刘虹；

【导师】 刘泉；

【作者基本信息】 武汉理工大学 ， 信息与通信工程， 2008， 博士

【摘要】 基于网络的信息交换为数字多媒体作品的使用、传播提供了便利的途径,同时由于数字作品极易被非法复制和篡改,使得数字作品的信息安全和版权保护成为迫切需要解决的问题。数字水印技术为数字化产品的版权保护提供了一种有效的方法。然而当前数字水印的研究主要集中在图像数字水印技术上,对视频和音频的数字水印研究相对较少。近年来,数字视频已成为大众生活中不可或缺的娱乐方式,而相应的版权保护技术尚未发展成熟,这就使得以视频数字水印为重要组成部分的数字视频产品版权保护技术的应用研究更为迫切。本文深入研究了数字视频版权保护和内容认证技术,设计了三种用于视频版权保护和内容认证的水印算法,并构建了基于Multi-Agent和数字视频水印的数字视频作品版权保护与内容认证系统,用于开放式网络环境下视频数据的防拷贝和版权实时跟踪与检测。本文主要研究工作如下:(1)针对电子商务中数字视频作品的盗版问题,研究了一种基于H.264压缩标准的鲁棒盲检测视频水印算法,用于视频数据的版权保护。算法在H.264帧内预测变换阶段,通过一定的判定条件自适应嵌入水印,将扩频处理后的水印序列嵌入到I帧每个宏块中所有4×4子块亮度分量的某个固定的高频DCT变换系数中,水印嵌入具有一定的自适应性,并在熵解码阶段实时盲提取水印,同时引入基于Lagrangian优化算法的编码控制,从而提高水印视频码流率失真性能,最后在H.264软件参考模型JM9.2基础上对水印的嵌入和提取进行了程序仿真实现,并对该算法进行各种实验测试和分析,结果表明,该算法在得到视频较好率失真性能的同时,具有高的鲁棒性,较好解决了数字视频作品的版权保护问题。(2)对于涉及到国家安全、法庭举证、历史文献等重要视频数据,如果出现非法篡改,将会造成不良的社会影响或重大的政治、经济损失。从视频对象内容完整性和可靠性认证出发,设计了一种视频内容认证半脆弱双水印方案。分别提取视频对象的时域特征值和空域特征值作为双水印信号,其中时域水印是取自H.264视频中I帧在时间轴上的序号,空域水印利用I帧亮度分量块的DCT系数绝对平均值得到。为了提高水印的安全性,对空域水印进行魔方变换加密。在视频内容认证过程中,首先对时域攻击进行认证,采用单向链表检测时域篡改类型。然后利用提取的空域水印和重构的空域检验水印对空域篡改进行定位。在实验仿真部分,分别从不可见性、时域认证、空域认证和实时性方面做了大量的实验,利用标准视频序列进行PSNR值计算、时域帧序号插入,删除,移位、空域内容删除,移动等多种测试,结果表明,本算法对视频合法操作具有鲁棒性,而对于非法操作具有脆弱性,可以对视频内容从时域和空域同时进行完整性认证,和对恶意篡改进行高精度定位。(3)在视频作品的多著作权问题上,需要多个不同的水印信号进行多版权认证,达到对数字视频作品在不同阶段的版权保护目的。在分析了单水印的局限性和多水印优势的研究基础上,设计了一种用于视频多版权保护水印算法。该算法首先用m序列变换对水印信号进行预处理,再提取视频数据的语义特征作为特征矩阵,最后将特征矩阵与水印信号结合生成一个版权矩阵,并把这个版权矩阵作为一个密钥,向可信任第三方(TTP,Trusted Third Party)注册,来保护自己对原视频作品的所有权。如果发现侵权时,通过这个密钥提取视频数据中与之对应的水印信号,来进行版权保护。并且各个水印之间相互独立,结果表明,该算法解决了视频作品多版权问题以及数字产品在发布、销售、使用等不同阶段的版权保护问题。(4)研究了开放式网络环境下数字视频作品版权管理系统框架及其存在的问题,然后介绍了数字水印与Agent在数字版权管理系统中的应用现状,构建了基于Multi-Agent和数字水印技术的数字视频版权保护和内容认证系统模型,适用于开放式网络环境下数字多媒体作品的安全资源共享。并对系统协议作了详细分析,最后通过软件实现了系统主要功能模块。结果表明该系统具有可靠性和实用性。更多还原

【Abstract】 The information exchange based on networks makes the application and transmission for digital multimedia works more convenient.However the information safety and copyright protection for digital works are becoming the most urgent problem to be solved because of the easy illegal copy and tampering.Now digital watermarking technology supplies an effective method to protect digital works copyright.At present,the digital watermark research is focused on image,but video and audio are less.In recent years,digital video has turned into an indispensable entertainment way,but the development of its corresponding copyright protection technology is unadvanced yet.Thus the digital video copyright technology is cried for more deep application study.Therefore,in this dissertation,the digital video watermarking technology for copyright protection and content authentication are made a deep study.Firstly three different watermarking algorithms are presented, which are employed to protect digital video copyright and validate video content. Then a system for digital video copy protection and-real-time copyright tracking in network is proposed.The main researches in this dissertation are illustrated as follows:(1) A robust and blind video watermarking scheme is presented to protect video copyright,which is based on H.264 video compression standard.According to the adaptive judgment condition,the spread frequency watermarking sequence is embedded into some special DCT coefficients of each I frame 4~*4 luminance blocks in the interframe prediction stage of H.264.Consequently,watermarking method is adaptive.In the entropy decoding process,watermark is examined blindly.Moreover, coding controlling is adopted,which is based on Lagrangian optimization algorithm. Thus the video bits stream rate distortion is improved.Finally,watermark embedding and extracting scheme are simulated via the software reference model for H.264—JM9.2 and many kinds of experiments are utilized.Experimental results show that the amended watermarking algorithm is imperceptible,real time,robust and safe for video copyright protection.(2) A semi-fragile dual watermark approach is proposed to authenticate video content.The time and space domain characteristic values of video are extracted to be the dual watermarks.The time domain watermark is the serial numbers of I frames and the space one is the absolute average value of DCT AC(Alternating Current) coefficients in I frame luminance block.In order to improve the watermarking safety, the space domain watermark is encrypted by magic cube transformation.In the process of video content authentication,time domain attacks are inspected firstly and the juggled type is examined through a unilateralism linked list.Then the space domain tampering is located according to the extracted watermark and reconstructed examining watermark.Finally,several standard video sequences are tested by common attacks,such as PSNR value,inserting/removing/cutting frame serial number,deleting/removing space content etc.Experiment results are completed to reveal the imperceptibility and real time of presented algorithm.(3) Limitation of single watermark and predominance of multi-watermark are introduced.Then a video copyright protection multi-watermark algorithm is researched in virtue of the concept of zero watermarking.At first,watermark is preprocessed by m-sequence transformation.Then the semantic characteristic of video data is extracted as its characteristic matrix.Finally,the characteristic matrix and watermark are combined to be an ownership matrix.Copyright owner could take this ownership matrix as a key to register it via Trusted Third Party to protect his right. Once piracy happens,the copyright owner could use the ownership matrix and the suspected video to distill the watermark to reveal his copyright.Different watermark is applied to generated different ownership matrix in each different stage,such as copyright,release,distribution and application etc.These watermarks are independent. According to experimental results,the proposed multi-watermarking scheme is good at solving video multi-copyright problem.(4) The unresolved problems of digital video copyright management system are discussed,and then the application of digital watermark and Agent in digital right management system is presented.Based on the previous research,a digital video copyright management system based on watermarking and multi-agent is proposed.It is suitable for safe resource sharing in open networks.Finally,the main function modules are realized by software programming.The system running shows that the proposed system is reliable and applied.更多还原