基于冗余小波变换的彩色图像水印算法研究

【作者】 渠长红；

【导师】 李会方；

【作者基本信息】 西北工业大学 ， 信号与信息处理， 2007， 硕士

【摘要】 数字水印技术是一种有效的数字产品版权保护和数据安全维护技术，是信息隐藏技术研究领域的一个重要分支。它主要包括水印生成、水印嵌入和水印的提取／检测三方面的内容。本文对小波域彩色图像水印技术进行了详细研究，给出一种水印置乱算法，对两种彩色图像水印算法进行了研究。本文的研究工作主要包括以下几个方面： 1．介绍了数字水印的需求背景、应用领域和国内外研究的现状，全面的总结了水印系统的概念、特性、分类、评测等相关问题。 2．讨论了水印技术的典型算法，并对它们进行评价，对空间域LSB、DCT域和DWT域水印算法进行了仿真。 3．对图像置乱技术的原理进行了分析，将置乱的图像作为秘密信息再进行隐藏，可以很大限度地提高水印算法的鲁棒性。在此基础上，给出一种算法简单、性能良好的置乱算法。 4．研究了一种基于离散小波变换(DWT)的彩色图像半盲水印新算法。该算法以有明确意义的二值图像作为水印，操作方便，运算量小，实时性好，提取时不需要原始图像。实验结果表明，该算法不但透明性好，而且对剪切、JPEG压缩、滤波等攻击有很好的鲁棒性。 5．给出了一种基于冗余离散小波变换(RDWT)的彩色图像水印算法。该算法将RDWT和奇异值分解结合起来。利用RDWT的冗余性和平移不变性这一主要特点，提高了水印算法的鲁棒性和嵌入容量。并根据奇异值分解在图像被施加小的扰动时其奇异值不会有太大的变化的性质，以及奇异值可以表征图像本身固有的一些代数属性的性质，使水印的稳健性和不可见性同时得到提高。算法实现了水印鲁棒性、不可见性和容量的统一。实验结果表明，该算法不但透明性好，而且水印嵌入量大，同时具有更强的鲁棒性，能很好地抵抗剪切、JPEG和JPEG2000压缩、滤波、噪声、旋转等攻击。本算法能适应各种形式的有意义数字水印的嵌入。更多还原

【Abstract】 Digital watermarking technology is regarded as the effective method to protect digital multimedia copyright and to ensure data security. It is one of the main parts of information hiding. Usually, a watermarking algorithm contains three parts: the design of watermark, the process of embedding and detection watermark. Based-wavelet color image watermarking is discussed in detail in this paper. An amended digital image scrambling method is proposed. Two color image watermarking algorithms are implemented. The two algorithms regard color image as carrier, and carry out significant image as watermark. The essential contents can be listed as follows:1. Introduce the need background of the digital watermarking, application field and the present condition of the domestic and international research, make a summary of the concept, characteristic, classification of watermarking.2. Discuss and evaluate the representative watermarking algorithms. Implement three algorithms based on LSB, DCT and DWT domain.3. The image scrambling theory is analyzed. Actually, scrambling is encrypting image. We pre-process the watermark by means of scrambling method, then embed encrypted image into carrier image. Image scrambling can improve the robustness of watermarking algorithms. Finally, an amended image scrambling method is proposed, and it is used by two nether color image watermarking algorithms.4. A novel semi blind color image watermarking algorithm based on DWT is devised. The algorithm uses foregoing scrambling method to encrypt significant binary watermark, extracting watermark without original image. The algorithm is simple and timesaving. Experimental results demonstrate that the algorithm is transparent and robust. It shows high robustness for the normal image processing methods such as cutting down, JPEG compression and filtering.5. A color image watermarking algorithm based on redundant wavelet transform (RDWT) is presented. The primary virtues of RDWT are redundancy and shift invariance. Robustness and capacity of algorithm can be enhanced by RDWT virtues. The main properties of SVD are as follows: the singular values of an image have very更多还原