【作者】 孟铁东；

【导师】 汤光明；

【作者基本信息】 解放军信息工程大学 ， 军事装备学， 2009， 硕士

【摘要】 视频隐写是信息隐藏近来的研究热点,各种隐写算法不断出现,加强视频隐写监控,确保视频隐写技术不被非法使用已经成为国家信息安全的重要组成部分。视频隐写检测是对视频隐写的攻击技术,它能解决视频隐写监控问题。本文对视频隐写检测技术进行了研究,主要研究内容包括:1.对视频隐写和隐写检测技术进行了深入分析。对视频隐写技术原理和特点进行了介绍,并对现有隐写算法进行分类,指出结合编码特点的隐写算法是未来发展方向。对视频隐写检测技术进行分类,并对现有的隐写检测现状进行总结,指出视频隐写检测急需解决的问题。2.针对扩频视频隐写检测,提出了基于帧重构的视频隐写检测算法。分析了扩频隐写原理并总结扩频隐写的特点,采用自适应邻域技术重构视频,计算待测视频与重构视频的差值并提取检测特征,最后设计了KNN分类器。实验结果表明,算法能有效地实现对扩频视频隐写检测。3.提出了针对运动向量隐写的检测算法。对运动向量隐写算法进行了分析,针对运动向量的时间变化,定义运动向量随机序列,将频率测试和游程测试得到的统计量作为运动向量帧内特征;针对运动向量的空间变化,定义运动向量的二值结构模式,将相邻帧的互信息作为运动向量帧间特征。最后基于运动向量的帧内和帧间特征,设计了针对运动向量隐写的检测算法,实验表明该算法能对运动向量隐写实施有效检测。最后,总结本文的工作,并对视频隐写分析技术的发展进行了展望,指出未来的研究方向和构想。更多还原

【Abstract】 Video data hiding has become the focus of information hiding recently, and kinds of algorithms have appeared. The supervisal to video steganography for preventing the illegal use of video steganography has been an important content of national security. Video steganalysis is the attack to the video steganography, which can solve problem of supervisal. In this paper, video steganalysis based on videos was discussed. The main contributions can be summarized as follows:1. Video steganography and steganalysis are investigated. The theory and characteristics of video hiding is introduced. Based on the classification of the existing video hiding algorithms, a conclusion that video coding would become the future orientation of video hiding is obtained. Then, video detection algorithms are classified, the actuality of which is concluded, and the problem in video steganalysis needed to be solved are pointed out.2. A video detection algorithm based on the frame reconstructure is proposed to detect the spread spectrum video data hiding. The theory and characteristics of spread spectrum steganography are analyzed and the technique of adaptive neighborhood is adopted to reconstruct the video. The difference between video and its reconstruct video is extracted as the feature for detection. Finally, KNN classifier is devised and the experiment is taken, the results of which shows that the algorithm can detect the video spread spectrum steganography effectively.3. A detection algorithm to attack the steganographic algorithm based on Motion vector (MV) is prevented. The steganographic algorithm based on MV is analyzed. Considering the time change of MV, the random sequence of MV is defined and the statistics got by frequency test and run test are used as the feature in frame of MV. Referring to the spatial change of MV, the mode of binary structure of MV is defined and the mutual information between neighbor frames is extracted as the feature between frames. Finally, based on the features extracted, a detection algorithm is devised and the experiment show the steganography based on MV can be detected effectively.Finally, the work of this paper is concluded. The further research topics about video hiding detection technique are outlined, and the research directions in future are discussed.更多还原