变换域自适应数字图像鲁棒水印算法研究
本文关键词: 离散小波变换 离散余弦变换 自适应 透明性 鲁棒性 出处:《江西理工大学》2017年硕士论文 论文类型:学位论文
【摘要】:信息时代的到来,计算机网络技术、多媒体信息技术以及相关学科的快速发展,使得数字产品被广泛应用开来,数字化信息在现代生活中变得愈加普遍。数字多媒体作品便于存储、复制及传播,藉此诱发的违法拷贝和版权盗用等信息安全问题愈渐增加,带来严峻的社会问题。因此,保障信息传输速度的同时,仍须保证个人信息、商业机密、知识产权等不被窃取。在版权保护领域中,数字水印占据着重要地位,通过把含有水印信息的标记隐藏到多媒体数字作品中,用以保护作品安全。但数字水印中鲁棒性和透明性两者彼此制约,在透明性得到保障的情形下,尽可能地提升其鲁棒性尤为关键。本文为了解决传统数字图像水印中版权保护的不足,提高数字作品的安全性,通过改进算法以均衡数字水印的透明性与鲁棒性,加强数字水印对各种攻击的抵御性能。本文的主要内容和创新点如下:⑴系统地论述了数字水印的发展背景和研究现状,分析了相关基础知识,研究了图像水印算法的相关设计手段,为本文的算法设计提供了充足的理论铺垫。⑵为了提升水印鲁棒性,均衡透明性与鲁棒性彼此之间矛盾,提出一种融合SVD和HVS优良特征的小波域数字图像水印算法。该算法首先对载体图像进行两级DWT,提取所得第二级低频子带并作剪裁处理;然后对每个子块进行SVD,并根据HVS对平均亮度的敏感性,结合修正的余弦相似性度量方法计算出水印的最佳嵌入位置;最后采取量化的形式把Arnold扰乱加密处理后的水印嵌入到选定奇异值中,其强度依照载体图像子块的奇异值自适应地调节。实验结果表明,该水印不仅具备良好透明性,对噪声、滤波、剪切以及JPEG压缩等水印攻击也都能够呈现出优异的鲁棒性。⑶针对水印透明性与鲁棒性彼此制约问题,提出一种融合DCT变换和LDPC码的数字图像水印算法。该算法首先对二值水印图像进行Arnold扰乱加密,之后再采取LDPC码对加密水印进行纠错编码;然后对载体图像按88?尺寸作剪裁处理,并分别对所得各个子块进行DCT变换;最后通过动态修改选定区域坐标处中频系数的大小关系来完成水印嵌入。实验结果表明,水印具备良好的安全性和透明性,能够有效地抵御滤波、JPEG压缩、噪声、旋转、剪切、平移、Gamma校正、直方图均衡化以及缩放等水印攻击,实现了水印的自适应嵌入和盲提取。
[Abstract]:With the advent of the information age, the rapid development of computer network technology, multimedia information technology and related disciplines, digital products have been widely used. Digital information is becoming more and more common in modern life. Digital multimedia works are easy to store, copy and disseminate, thus causing more and more information security problems, such as illegal copy and copyright embezzlement, which bring severe social problems. While guaranteeing the speed of information transmission, we must also ensure that personal information, trade secrets and intellectual property rights are not stolen. In the field of copyright protection, digital watermarking plays an important role. In order to protect the security of digital works, the watermark information is hidden in the multimedia digital works. However, the robustness and transparency of digital watermarking are restricted to each other, and the transparency is guaranteed. In order to solve the shortcomings of copyright protection in traditional digital image watermarking and improve the security of digital works, the algorithm is improved to balance the transparency and robustness of digital watermarking. The main contents and innovations of this paper are as follows: 1: 1 systematically discusses the development background and research status of digital watermarking, and analyzes the relevant basic knowledge. This paper studies the relevant design methods of image watermarking algorithm, and provides sufficient theoretical basis for the algorithm design in this paper. 2. In order to improve the robustness of watermark, the contradiction between transparency and robustness is proposed. A digital image watermarking algorithm in wavelet domain is proposed, which combines the excellent features of SVD and HVS. Then the best embedding position of the watermark is calculated according to the sensitivity of HVS to average brightness and the modified cosine similarity measurement method. Finally, the Arnold scrambled watermark is embedded into the selected singular value in the form of quantization. The intensity of the watermark is adjusted adaptively according to the singular value of the sub-block of the carrier image. The experimental results show that the watermark not only has good transparency, but also has good noise. Watermarking attacks such as filtering, clipping and JPEG compression can also show excellent robustness. 3. 3 constraints on the transparency and robustness of the watermark. A digital image watermarking algorithm combining DCT transform and LDPC code is proposed. Firstly, the binary watermark image is scrambled by Arnold, then the encrypted watermark is coded by LDPC code, and then the carrier image is coded by 88? The dimension is clipped, and each sub-block is transformed by DCT transform. Finally, the watermark is embedded by dynamically modifying the relationship of the intermediate frequency coefficient at the selected region coordinates. The experimental results show that, The watermark has good security and transparency, and it can effectively resist the filtering attacks such as JPEG compression, noise, rotation, cutting, translation Gamma correction, histogram equalization and scaling, which realizes the adaptive embedding and blind extraction of the watermark.
【学位授予单位】:江西理工大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TP309.7
【参考文献】
相关期刊论文 前10条
1 唐振华;梁聪;区骋;黄旭方;覃团发;;基于DCT域纹理特征的多聚焦图像融合[J];应用科学学报;2015年06期
2 丰明坤;赵生妹;施祥;;视觉多通道梯度与低阶矩自适应图像评价[J];仪器仪表学报;2015年11期
3 刘真;白韬韬;卢鹏;;一种解密图像无背景噪声的加密全息数字水印技术[J];光学学报;2015年02期
4 邱应强;余轮;;基于整数变换的图像可逆水印方法[J];电子测量与仪器学报;2015年01期
5 姜传贤;杨铁军;董明刚;程小辉;李智;;基于线性空间隐藏模型的可逆图像水印算法[J];自动化学报;2014年10期
6 冷亚军;陆青;梁昌勇;;协同过滤推荐技术综述[J];模式识别与人工智能;2014年08期
7 朱新山;陈砚鸣;董宏辉;丁杰;;基于双域信息融合的鲁棒二值文本图像水印[J];计算机学报;2014年06期
8 苑玮琦;刘笑楠;孙晓;滕海;;一种虹膜图像块状纹理检测算法[J];仪器仪表学报;2014年05期
9 吴成茂;;离散Arnold变换改进及其在图像置乱加密中的应用[J];物理学报;2014年09期
10 王新宇;詹永照;;构造顶点分布特征的三维模型数字水印算法[J];计算机辅助设计与图形学学报;2014年02期
相关硕士学位论文 前1条
1 牛海静;基于DWT和SVD的多重鲁棒数字水印算法[D];华南理工大学;2013年
,本文编号:1538979
本文链接:https://www.wllwen.com/falvlunwen/zhishichanquanfa/1538979.html