基于脆弱水印的H.264/AVC视频认证技术研究

Research on H.264/AVC video authentication based on fragile watermarking

摘要:(摘要内容经过系统自动伪原创处理以避免复制,下载原文正常,内容请直接查看目录。)

跟着互联网技巧和盘算机技巧的成长和以H.264/AVC编码的视频的普遍运用,人们急需对H.264/AVC视频的内容完全性做认证。软弱水印技巧不许可认证对象有任何一点的改动,可以用来对内容完全性认证。所以,研讨H.264/AVC的软弱水印技巧异常成心义。本文提出并改良了两种用于H.264/AVC内容认证的软弱水印算法,重要任务以下:第一种算法,将由Ⅰ帧重建帧的Hash值用流暗码加密后,生成的一个具有同步功效的水印信息,嵌入到以后GOP组的P帧和B帧当中。嵌入办法的根本思惟是依据嵌入信息修正16×16宏块的拖尾系数的个数,联合田和赵等人的算法可以或许完成每嵌入5比特水印信息至少修正一个4×4子宏块的拖尾系数。试验成果注解,将水印信息嵌入到P帧和B帧,可以或许削减掉真漂移;本算法具有对Ⅰ帧完整级的掩护才能,同时还具有对帧删除,帧增长的辨别才能,对P帧和B帧的内容改动具有必定的改动定位才能;嵌入水印后其实不会给视频带来较年夜的视觉掉真和比特率增长。为了加强对P帧和B帧的认证才能和防止掉真漂移。本文又提出了一种H.264/AVC的可逆软弱水印算法。该算法起首基于直方图特点函数质心道理盘算一组滤波前重建视频帧的特点,然后把由特点、装备编号和时光戳生成的独一认证水印信息,经由过程可逆的算法嵌入到下一个Ⅰ帧亮度重量残差的量化后DCT系数上。嵌入算法的根本思惟是把待嵌入4×4子宏块的系数个数修正为地点宏块中最多,使之嵌入水印后具有两个特色之一:一是系数个数最多且最初一个非零系数为-1,二是最初一个系数为一切宏块最初一个系数中相对值最年夜。验证时,只需把从待验证视频中提取的认证信息与重构的认证信息停止比对就可以完成认证。试验成果注解,本文提出的算法可以或许胜利地嵌入和提掏出水印信息并准确地恢复系数;算法对帧内内容改动、帧删除、帧拔出、帧交流、重紧缩、惯例旌旗灯号处置等改动异常敏感;另外,嵌入容量年夜,同时对视频的质量和紧缩比特率影响较小,可以或许对视频帧的一切色彩重量停止完全性认证。

Abstract:

Along with the Internet technology and computer technology development and widespread use of video encoding to H.264/AVC, there is an urgent need to complete the H.264/AVC video content authentication. Weak watermark techniques do not license any object changes, can be used for content authentication. Therefore, the study of H.264/AVC weak abnormal meaningful watermarking techniques. In this paper, and the improvement of the two for h.264/avc content authentication of the weak watermark algorithm, an important task is as follows: the first algorithm, by the I frame reconstruction frame hash value using a stream cipher encryption and generated a watermark information with a synchronous effect and embedded into the group of GOP P and B frames which. Embedded in the way of the fundamental ideas are based on the number of embedded information correction 16 * 16 macro block of the trailing coefficient, combined field and Zhao et al algorithm can probably through embedded in every 5 bits watermarking information at least modify a 4 x 4 sub macro blocks of the trailing coefficient. Test results of the annotation, watermarking information is embedded in the P and B frames, you can probably cut really drift; the algorithm is of I frame integrity level of cover to. At the same time, also has in frame deletion, frame growth ability to identify, the contents of the P frame and B frame changes with certain changes made to; after embedding watermark actually not to video bring more Nianye visual off it and bit rate of growth. In order to strengthen the P frame and B frame to prevent distortion drift and authentication. This paper proposes a reversible H.264/AVC weak watermarking algorithm. Chapeau of the algorithm based on histogram characteristic function principle of centroid calculation as well as the filtering reconstruction characteristics of video frame, then the generated by the characteristics, equipment number and time stamp only authentication watermarking information, via the reversible process algorithm is embedded into the DCT coefficients after quantization of the next I-frame brightness weight residuals. Embedding algorithm is the fundamental thinking is the correction to be embedded in 4 * 4 sub macro blocks of the coefficient number to place the macro block most, the embedded watermark is one of the two characteristics: A is the number of coefficients most and was originally a non-zero coefficient is - 1, the second is initially a coefficient of all macroblock initially a coefficient of relative value of most of the eve of the. The test, only the authentication information extraction and reconstruction from video to verify the authentication information on stop can be completed certification. Annotation of results, algorithm is proposed in this paper can or Xu victory embedded and extract the watermark information and accurately recover coefficient; algorithm for intra frame content to modify, delete frames, pulled out of the frame, frame exchanges, heavy tight, practice signal disposal changes abnormally sensitive; in addition, embedding capacity Nianye, also on the video quality and compression bit rate has little effect, can perhaps the video frames of all colors weight stop for authentication of the.

目录:

摘要6-7
Abstract7-8
第1章 绪论11-17
    1.1 研究背景和意义11-12
    1.2 国内外研究现状12-15
    1.3 本文主要工作和安排15-17
第2章 H.264/AVC脆弱水印技术17-31
    2.1 H.264编码标准17-21
        2.1.1 相关术语及约定17-20
        2.1.2 H.264编解码器结构20-21
    2.2 H.264脆弱水印技术简介21-22
    2.3 H.264关键技术点及其脆弱水印技术22-27
        2.3.1 帧内预测23-24
        2.3.2 帧间预测24
        2.3.3 变换和量化24-26
        2.3.4 熵编码26-27
    2.4 H.264脆弱水印主要攻击形式27-28
    2.5 脆弱水印性能评价28-30
        2.5.1 主观质量评价28-29
        2.5.2 客观质量评价29
        2.5.3 比特率影响29-30
    2.6 总结30-31
第3章 基于残差DCT系数的脆弱水印算法31-42
    3.1 算法概述31
    3.2 水印生成及嵌入算法31-33
    3.3 提取算法33-34
    3.4 实验结果及分析34-41
        3.4.1 主客观质量评价35-37
        3.4.2 比特率影响37-38
        3.4.3 认证能力分析38-40
        3.4.4 对比实验40-41
    3.5 结论41-42
第4章 基于直方图特征函数的可逆脆弱水印算法42-59
    4.1 算法概述42-43
    4.2 水印生成算法43-46
        4.2.1 脆弱特征生成算法43-45
        4.2.2 像素值置乱方式45
        4.2.3 认证水印生成算法45-46
    4.3 水印嵌入46-47
    4.4 水印提取47-49
    4.5 认证过程49-50
    4.6 实验结果及分析50-58
        4.6.1 脆弱性分析50-52
        4.6.2 主客观质量评价52-54
        4.6.3 比特率影响54-56
        4.6.4 算法可逆性分析56-57
        4.6.5 对比实验57-58
    4.7 结论58-59
总结与展望59-60
致谢60-61
参考文献61-65
攻读硕士学位期间发表的论文65