【作者】 刘镔；

【导师】 刘粉林；

【作者基本信息】 解放军信息工程大学 ， 计算机软件与理论， 2008， 博士

【摘要】 由于信息隐藏技术能够在一定程度上解决传统密码学在某些应用领域中的不足,因而其被视为信息安全研究领域的新鲜血液,成为数字化信息时代有效应对信息安全挑战、维护网络信息安全的主要手段之一。现有信息隐藏研究的载体对象主要以静态图像为主,对视频信息隐藏技术的研究目前尚未完全展开。但是随着流媒体技术的发展,数字音、视频资源正逐步取代文字和静止图像成为网络信息的主流,视频信息隐藏技术日渐成为研究热点之一。本文主要围绕下列问题开展研究:1、基于共谋策略的视频空域隐写分析:针对视频序列中存在的局部运动严重干扰隐写分析精度的问题,提出了一种基于帧间共谋策略的视频隐写分析框架。该框架将秘密信息和局部运动建模为双模噪声,首先基于视频帧间分块相关度量的策略提取特征,通过分类方法判别出可疑视频帧,然后将可疑帧提取为静态图像,应用相对更加成熟的图像隐写分析算法实施进一步的检测,以期最终获得较高的检测精度。2、压缩视频VLC码字中的实时信息隐写:针对视频压缩域中现有隐写算法的隐写容量受编码格式限制的问题,提出了一种基于视频VLC码对映射的实时隐写算法。该算法在保证载体视频码率不变的前提下提高了秘密信息的嵌入容量,信息的嵌入和提取操作完全在VLC码字中进行,不需要对载体视频进行全部或部分的解压和重压缩,从而能够满足实时性要求。3、基于优化算法的压缩视频安全隐写:针对视频隐写算法设计中因忽略视频时域特性而导致安全性下降的问题,对安全视频隐写的算法设计准则进行了初步探讨,基于全局优化策略给出了兼顾时域、空域特性的压缩视频安全隐写框架,能够依据实际的应用需求调整目标函数,以确定适当的信息嵌入方案。4、基于内容特征的压缩视频分层认证:针对压缩视频内容真实性认证问题,提出了一个基于压缩视频内容特征的分层认证方案。该方案结合视频帧间和帧内的内容特征可对其内容真实性进行不同层次的认证。方案所选取的内容特征具有半脆弱的特性,即对于恶意操作较为敏感,而对正常的处理操作比较鲁棒,基于特征的半脆弱特性可以区分针对视频所进行的正常处理操作和恶意内容篡改。认证方案中,首先以较小的计算量快速检测恶意篡改是否发生,当发现恶意篡改后,能够根据需要进一步确定视频被篡改的具体位置。5、基于数字指纹的盗版追踪:针对数字视频等多媒体产品发行中经常遭遇的盗版问题,给出了一类基于抗共谋攻击数字指纹的盗版追踪、取证方案。在方案中给出了基于可信第三方的数字视频作品分发协议,该协议的设计兼顾了视频作品网络发布的实用性和安全性;在方案中的指纹编码方面,通过构造性证明方法证明了一定共谋尺度下完全抵抗各类共谋攻击的数字指纹编码方案的存在性;同时给出了基于严格正交指纹编码的抗平均共谋指纹编码方案以及当平均共谋发生时能够准确发现共谋参与者的树形追踪算法。最后,对全文做了总结,并对下一步的研究工作进行了展望。更多还原

【Abstract】 As a remedy for the traditional cryptography technology, information hiding plays a more and more important role in the information security research, and becomes the main countermeasure to the network information security threats in the digitalize age. In its infant times, information hiding algorithms are designed with the still image cover commonly. However, with the rapid development of stream media technology, digital audio and video resources are taking the place of text and image to be the dominant information format in present networks. Therefore, video information hiding technology becomes the hotspot of information technology research. The main works of this thesis are as the following:1. Collusion-based video space domain steganalysis: the existence of local motion will disturb the detecting precision of present video steganalysis algorithms. An interframe collusion based video steganalysis framework is proposed for higher hiding message detection exactness. The video steganalysis process is divided into two steps. In the first step, local motion and message embedded in video is treated as a bimodal noise. Suspicious video frames will be recognized by decision module employing features extracted with inter-frame correlation. In the second step, suspicious frames will be analyzed critically by the present powerful image steganalysis schemes.2. Compressed video real time steganography in the VLC codes: When embedding message in the compressed video, the message capacity will be limited by the compress coding standard. A real time video staganography algorithm based on VLC code pairs mapping is proposed. With the less change of compressed video code rate, the capacity of message will be increased by using the proposed algorithm. Moreover, the embedding and extracting operation are wholely performed in the VLC codes, with no need for full or even partial decompression for the real time requirement.3. Optimal strategy on compressed video secure steganogrphy: Present frame by frame approaches to video steganography will bring a serious security pitfall. Temporal dimension is a crucial point in video and it should be taken into account to design efficient algorithms. With the highlight of perceptual shaping research, an optimal strategy on compressed video secure steganography is proposed. With the optimal strategy, an appropriate message embedding scheme is determined in a closed-loop manner with a cost function for minimization, to enhance the antisteganalysis capability and image quality after data embedding.4. Content feature based compressed video hierarchical authentication: For the compressed video authentication, a content feature based hierarchical authentication scheme is proposed. The scheme combined the inter-frame and intra-frame features to ensure the authenticity of video content with two different level. Features selected have semi-fragile property, with which can distinguish malicious attacks such as erasure, insertion and replacement from video processing. The proposed scheme can detect the existence of malicious attacks with a light computer payload and further locate the corrupt areas.5. Anti-collusion digital fingerprint based traitor tracing: How to defense collusive attack is one of the key problems of copyright protection of digital contents. An anti-collusion fingerprint based traitor tracing scheme is proposed. The tracing protocol is based on the trusted third party, which concerns on both security and practicality. The existence of complete anti-collusion fingerprint scheme is proved in a construct manner. The fingerprint coding scheme and corresponding tracing protocol are proposed by using an orthogonal sequence to construct the fingerprint bits. Under the average collusion attack, by the fingerprint algorithm, the owner can trace the traitors quite efficiently.At last, the research work in this thesis is summarized and potential research topics in future are discussed.更多还原