【作者】 代江艳；

【导师】 吕英华；

【作者基本信息】 东北师范大学 ， 计算机软件与理论， 2010， 硕士

【摘要】 随着计算机网络的应用推广以及数字多媒体技术的迅速发展,视频在传输过程中遭到攻击或侵权的问题越来越严重。目前,视频信息隐藏作为信息安全的重要手段之一,已经成为信息隐藏方向研究的一个重点问题。本文首先简单介绍了视频信息隐藏的研究现状和发展趋势,然后简要介绍了视频信息隐藏的基础知识,包括视频信息隐藏的分类、特点和攻击类型以及本文用到的相关数学理论和方法,最后提出了两种视频信息隐藏方法。一种是基于帧间差分和离散小波变换的视频水印方法,该方法结合人类视觉系统(HVS)的特性,使用基于块的帧间差分运动分析方法对经过小波变换后的视频序列进行运动分析,将Arnold变换后的二值水印图像嵌入到得到的运动活跃区域。另外一种是在研究视频水印的基础上,提出的一种基于生物认证图像集的视频密写方法,该方法将生物认证技术与信息隐藏技术相结合,首先对视频序列和生物认证图像集进行相关性分析,降低秘密信息的能量,将得到的残差秘密图像集作为待嵌入的秘密信息;然后采用基于帧的和基于块的帧间差分方法对视频做运动情况分析;最后将秘密信息嵌入到运动比较活跃的区域。本文使用MATLAB作为实验平台,针对提出的视频水印和视频密写方法进行了大量实验,得到了精确的实验数据并进行了分析。实验结果表明,本文提出的两种视频信息隐藏方法都具有很好的不可见性、鲁棒性和安全性。另外,本文提出的视频密写方法充分利用了视频序列的冗余信息,实现了大容量秘密图像集的嵌入,提高了生物认证的识别率,为进一步的研究奠定了基础。更多还原

【Abstract】 In recent years, with the wide applications of Internet and the rapid development of digital multimedia technology, the problems of suffering from attacks and the infringement of copyright in the transmission of videos are become more serious. At the moment, as one of the most powerful tools for information security, the video information hiding has been an important problem for the research of information hiding.In this paper, firstly we give an outline of the situation of study and the development of information hiding. Secondly, we summarize the rudimentary knowledge, including the category, characteristics of video information hiding and types of attack. And we briefly introduce the related mathematical theories and methods. At last, we propose two video information hiding schemes. One is a video watermarking approach based on temporal difference and discrete wavelet transform. It combines with human visual system(HVS) characteristics, uses block-based temporal difference to analyze the motion information of approximate bands after discrete wavelet transform decomposition of video frames and finally embeds a binary watermark scrambled by Arnold transform to obtained motion activity regions. And the other is a video steganography approach based on biometric image set on the basis of video watermarking research. This method tries to combine biometric identification technology with information hiding technology. Firstly it analyses the correlation between the biometric image set and the video sequence for reducing the power of secret images and takes the obtained residual images as secret information which will be used to embed. And then it adopts frame-based and block-based temporal difference to analyze the motion activity of video sequences. Finally we embed the secret information to motion regions which have higher activity.For the proposed video watermarking approach and video steganography approach, we take MATLAB as the platform to conduct many tests, gain many exact experimental results and analyze them. Extensive experimental results demonstrate that the proposed methods have good invisibility, perfect robustness, and high security. Besides, the proposed video steganography approach makes good use of the redundancy information, implements the high capacity embedding of secret image sets, improves the biometric recognition rate, and lays the foundation of further research.更多还原