一种鲁棒性佳的视频水印算法

VIP免费

3.0 陈辉 2024-11-19 4 4 1.4MB 47 页 15积分

侵权投诉

摘要

视频水印涉及多媒体领域的版权保护、数字认证、完整性检测等内容，是信

息隐藏的一个重要分支。网络和计算机技术的发展使得多媒体传播和存储方式产

生了极大变化，也使得多媒体的版权保护和数字认证变成难题，因此视频数字水

印具有重要的研究意义。视频水印可以分为基于原始视频的水印和基于压缩域水

印，对原始视频水印的研究可以继承图像水印的部分发展成果。由于视频信息具

有信息量大的特点，在传输和存储之前需要对其进行压缩，由此产生的对视频信

息压缩域水印的研究就具有较高的价值。

本论文首先分析了视频水印的分类及几种基本的水印嵌入算法，然后针对基

于场景分割的几种方案做了研究和评析，并分析了压缩域的基于纹理信息的水印

嵌入方案。在此基础上，本论文提出了具有良好鲁棒性的基于原始视频和压缩域

视频的水印嵌入和提取算法。

文中对水印信息的预处理包括二维 Logistic 混沌加密和 Turbo 码编码，利用混

沌映射的非周期性置乱和对密钥极其敏感的特性提高视频水印的安全性。将对水

印的攻击视作信道噪声，

Turbo 码可以对攻击过的水印信息进行纠错处理，从而提

高水印的鲁棒性。

针对原始视频水印，考虑视频流的时间轴特性，通过计算相邻帧的差值并采

用双阈值区分场景，这样可以克服渐变场景难以判断的情况。在不同的场景嵌入

水印的不同部分，不同部分水印以索引信息区分，增加了水印对时间轴的抗攻击

能力。通过对视频帧进行三级小波变换，并分块进行离散余弦变换，使得水印透

明型好。实验证明，该方案嵌入提取效果良好，可以有效抗普通攻击，以及帧移

位、帧丢失、帧交换等攻击。

在压缩域，本文根据视频的纹理特性进行水印的嵌入和提取。根据视频宏块

的背景亮度、纹理复杂度等特性，对宏块信息进行基于人类视觉系统的分类，然

后利用分类的信息生成掩蔽矩阵，从而确定嵌入水印的宏块位置。把 4×4 的DCT

块数据分为直流分量、轮廓分量和高频分量等，每个不同的分量嵌入水印信息的

权重不同，从而更有利于水印嵌入的掩蔽性。实验表明该算法具有较强的鲁棒性。

关键词：视频水印场景分割压缩域 DCT

ABSTRACT

The video watermarking is involved in the field of multimedia copyright protection,

digital authentication, integrity checking etc., and it is an important branch of

information hiding. Network and computer technology make multimedia

communication and storage changed greatly, but also make copyright protection and

multimedia digital certification become a difficult problem. So the video watermarking

researches are very important. The researches on the watermarking in the original

videos can be based on some researches on the watermarking in the images. Videos

should be compressed before transmission and storage because of a huge amount of

information in them. The researches on compressed domain watermarking are also very

important.

The classification of video watermarking and some basic algorithms are proposed

in details in the paper. Some researches and assessments are made based on scene

segmentation for several programs, and the paper analyzes the compressed domain

watermarking based on texture information. This paper summarizes the results of

previous researches based on the original video and compressed domain video

watermarking algorithms and proposes a good robust watermarking scheme.

In this paper, the preprocess on watermarking includes two-dimensional Logistic

chaotic encryption and Turbo coding. We advance the security on video watermarking

by the a periodic scrambling on the chaotic maps, which are also very sensitive to the

security key. The attacks can be thought as watermark channel noises, and Turbo codes

can correct the attacked watermarking, which improve the robustness of the watermark.

For the original video watermarking, considering of the timeline feature on the

video streams, we use two thresholds distinguishing program scenes, which can

overcome the difficulties on the judgment of the gradients. To advance the capability of

the watermarking on the anti-attack to the timeline attack, we embed watermarks in

different scenes with different parts, and the different parts of watermarks have different

indexes. By the three wavelet transformations on the video frames and discrete cosine

transformations blocks by blocks, we make the watermarking with the general

advantages in frequency domain, such as high transparency, favorable combination of

the visual system to improve shelter and so on. Experiments show that this kind of

watermarking has a good effect on the embedding and extracting, and it is resistant to

common attacks, frame shift, frame loss and frame exchange attacks.

In compression domain, the paper analyses the main texture characteristics and

designs the video watermarking embedding and extraction algorithm. The paper

classifies macro blocks in the videos based on human visual system, the classification

standard referent to the background luminance, complexity of texture, and generate

masking matrixes with the classified macro blocks to determine the location of the

macro blocks where should be embedded watermarks. Within the macroblock of 4 × 4

III

data classified as DC component, contour component, high frequency components and

so on. For each of the different components of the watermark information embedded in

different weights. Thus more conducive to the masking of the watermark embedding.

Experiments show that the algorithm has strong robustness.

Key Word: Video watermark, Scene segmentation, Compressed

Domain, DCT

中文摘要

ABSTRACT

第一章绪论.................................................................................................................1

1.1 数字水印研究背景及意义........................................................................1

1.2 国内外研究的现状以及数字水印面临的难题........................................2

1.3 论文结构、章节安排和主要研究成果....................................................6

第二章视频水印的理论基础.........................................................................................7

2.1 视频水印的性能要求................................................................................7

2.2 视频水印的分类........................................................................................8

2.3 场景分割..................................................................................................10

2.4 H.264 视频压缩技术................................................................................14

2.5 视频水印性能评价..................................................................................18

2.6 本章小结..................................................................................................20

第三章基于原始视频的数字水印算法.......................................................................21

3.1 基于原始视频水印采用的关键技术......................................................21

3.1.1 二维 Logistic 混沌加密方案 ....................................................... 21

3.1.2 双阈值场景分割策略...................................................................23

3.2 对水印信息的预处理..............................................................................25

3.3 算法实现的过程......................................................................................26

3.4 实验结果及抗攻击性评估......................................................................28

3.4.1 嵌入提取的效果和攻击测试.......................................................28

3.4.2 效果评估.......................................................................................30

3.5 本章小结..................................................................................................30

第四章基于 H.264 压缩标准的数字水印算法 ...........................................................31

4.1 水印信息预处理过程..............................................................................31

4.2 嵌入提取算法..........................................................................................32

4.2.1 嵌入位置确定 .............................................................................. 32

4.2.2 嵌入算法 ...................................................................................... 33

4.3 实验结果与评估 ......................................................................................35

4.4 本章小结 ..................................................................................................36

第五章研究总结与未来展望.......................................................................................37

5.1 论文总结 ..................................................................................................37

5.2 未来工作展望 ..........................................................................................37

参考文献.........................................................................................................................39

在读期间公开发表的论文和承担科研项目及取得成果.............................................43

一、发表论文.................................................................................................43

二、科研项目.................................................................................................43

致谢.............................................................................................................................44

边框:底端: (无框线)

设置格式[oempc]:

删除[oempc]:

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

15 积分 4人已下载

立即下载 VIP免费下载

摘要：

I摘要视频水印涉及多媒体领域的版权保护、数字认证、完整性检测等内容，是信息隐藏的一个重要分支。网络和计算机技术的发展使得多媒体传播和存储方式产生了极大变化，也使得多媒体的版权保护和数字认证变成难题，因此视频数字水印具有重要的研究意义。视频水印可以分为基于原始视频的水印和基于压缩域水印，对原始视频水印的研究可以继承图像水印的部分发展成果。由于视频信息具有信息量大的特点，在传输和存储之前需要对其进行压缩，由此产生的对视频信息压缩域水印的研究就具有较高的价值。本论文首先分析了视频水印的分类及几种基本的水印嵌入算法，然后针对基于场景分割的几种方案做了研究和评析，并分析了压缩域的基于纹理信息的水印嵌入方...

展开>> 收起<<

一种鲁棒性佳的视频水印算法.pdf

共47页,预览5页

还剩页未读，继续阅读

一种鲁棒性佳的视频水印算法

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

推荐作者

热门标签

举报选择: