【作者】 陈大庆；

【导师】 薛鸣球； 顾济华；

【作者基本信息】 苏州大学 ， 光学工程， 2010， 博士

【摘要】 计算全息和数字水印相结合的全息水印是一种信息隐藏新技术,由于全息图的不可撕毁性等特点,全息水印具有很强的鲁棒性,为保护图像、音频、视频等知识产权提供新的有效手段。本文以信息光学为理论基础,研究计算全息的数字水印新方法,以获得比国际上典型的计算全息数字水印法更高的鲁棒性,得到性能优良的全息水印新方案。论文研究了一种基于纯相位干涉的计算全息图像信息隐藏技术。分析了相位在变换中的重要性,并采用纯相位替代传统傅里叶变换全息中的频谱复振幅与参考光波发生干涉,从而得到高对比度的计算全息图。为提高安全性,采用密钥将全息图进行加密,并通过离散余弦变换在频域嵌入宿主图像中完成信息隐藏。信息提取时先用密钥对提取的全息图进行解密,然后完成光学或计算全息再现过程。理论分析和仿真实验表明,纯相位的傅里叶变换计算全息信息隐藏技术对有损压缩、剪切和滤波等多种图像处理操作均具有很高的稳健性,并且比传统傅里叶变换全息信息隐藏的稳健性有很大提高。论文提出了基于相位恢复和计算全息技术的图像水印方法。该方法采用相位恢复算法将需要隐藏的水印图像编码为纯相位,然后用该纯相位代替传统的傅里叶变换全息中的物光波频谱与参考光波发生干涉,得到理论对比度为100%的傅里叶变换全息图。理论分析和数值计算实验表明了改进后的方法具有更好的稳健性和实用价值。论文基于信息光学理论提出了一种非线性限幅傅里叶计算全息的数字水印方法,使数字水印具有更好的鲁棒性。在传统傅里叶变换全息基础上提取计算全息图的振幅和相位,将相位信息作为输入函数,振幅信息转换成偏置函数,经过非线性限幅处理后得到二元傅里叶计算全息图。由于二元全息图比普通全息图具有更强的抗噪性能,从而大大提高了数字水印的鲁棒性。为了提高水印的抗压缩能力,通过离散余弦变换在频域嵌入宿主图像中。理论分析和仿真实验证实了在适当降低了初始提取水印精度的情况下,该方法比传统傅里叶计算全息水印的鲁棒性有很大提高。论文提出了一种新的基于相位恢复和非线性限幅的图像信息隐藏技术。该方法采用相位恢复算法将秘密信息编码为纯相位,然后用平行波前作为载波,运用非线性限幅模型处理得到二元全息干涉图,并给出理论推导公式。该全息图比普通非线性限幅的二元全息图包含更多的原始图像信息,提高了再现波前的衍射效率,从而大大提高了该技术的鲁棒性。采用密钥将此全息图进行加密后,结合JPEG系统基本原理运用双极性量化算法在频域嵌入载体图像中完成信息隐藏。信息提取时不需要原始载体图像的参与,通过密钥将提取的全息图进行解密,再进行数字再现即可完成。理论分析和对标准测试图像实验表明,该技术对多种图像处理操作稳健性有很大提高,具有很高的实用价值。论文提出了一种在计算全息技术中引入相位恢复算法的音频水印方法。用水印图像经相位恢复处理形成虚拟物光波与参考光波干涉,得到高对比度的傅里叶变换全息图。利用量化算法和离散余弦变换把计算全息图嵌入音频信号中,在水印提取过程中不需要原始音频信号的参与,并且采用密钥加强了水印的安全性。数值计算实验表明:该算法对有损压缩、重采样、低通滤波、噪音干扰等常用音频信号攻击均具有很好的稳健性。更多还原

【Abstract】 CGH digital watermarking is a new information hiding technique that combined with computer-generated hologram and digital watermarking. The watermark has strong robustness because of non-tear of holograms, etc. It provides new effective means for protect intelligent property, such as image medium, audio medium and video medium, etc. We study the new approach of digital watermarking using CGH based on information optics theory, which in order to acquire a good project and higher robustness than the international typical method.We study the approach of image information hiding based on CGH of phase-only interference holography. The importance of phase in the transformation is analyzed. Instead of the complex amplitude of the traditional Fourier-transformed object wave, the phase-only wave is used to interfere with the reference wave, and a high contrast Fourier-transformed CGH is obtained. Hologram is embedded into host image by discrete-cosine-transform algorithm and encrypted by the security key. The original information is retrieved by reconstruction of optics or computer-generated hologram with the security key. Theoretical analysis and simulation experiments show that the algorithm has more robustness to image lossy compression, cropping and filtering, comparing with the algorithm based on general computer-generated hologram.In the next part of the dissertation, a new approach of digital watermarking based on phase retrieval algorithm and computer-generated hologram is proposed. The watermarking image is transformed into virtual object wave after phase retrieval algorithm. Then use the virtual object wave take the place of traditional Fourier- transformed object wave. It interfered with the reference wave to form Fourier- transformed computer-generated hologram and the contrast is 100% in theory. Theoretical analysis and numerical experiments show that the algorithm has more robustness and practical value. We proposed a new approach of Fourier CGH digital watermarking with nonlinear amplitude limiting based on information optics,which realizes better robustness. The amplitude and phase are extracted on the basis of traditional Fourier transform holography. Binary Fourier hologram was formed by nonlinear amplitude limiting algorithm when the phase information as input function and the amplitude information transferred into a bias function. Because the binary hologram has stronger anti-interference property than the normal hologram, the digital watermarking obtains much better robustness. To improve the watermark’s resistance to compression, hologram is embedded in the frequency domain through discrete cosine transform. Theoretical analysis and numerical experiments show that the algorithm has more robustness under the situation that reduce initial watermark accuracy within limits, comparing with the algorithm based on traditional Fourier CGH watermarking.A new approach of information hiding technology based on phase retrieval algorithm and nonlinear amplitude limiting is proposed. The secret information is encoded into a pure phase by phase retrieval algorithm, and using parallel wave as the carrier. Then we obtain binary hologram by nonlinear amplitude limiting model. The theoretical derivation formula is given. The hologram contains more original image information than the hologram getting from average nonlinear amplitude limiting model and improves the diffraction efficiency when reconstruction, and which greatly improves the robustness of the technology. Hologram is encrypted by security key and embedded into host image in the frequency domain by the basic principles of JPEG and bipolar quantization algorithm. Information extraction does not require the participation of original image. It can be extracted with the security key and the secret information is retrieved by digital reconstruction. Theoretical analysis and experiments on standard test images show that the algorithm has more robustness and practical value to a variety of image processing operations comparing with the algorithm based on general digital holography.A new approach of audio watermarking by use of phase retrieval algorithm and computer-generated hologram is proposed. The watermarking image is transformed into virtual object wave after phase retrieval algorithm. It interferes with the reference wave to form high contrast Fourier-transformed CGH. And the hologram is embedded into audio signal by quantification process and discrete-cosine-transform .The watermark can be extracted without the original audio signal and the security key provides a good security. Numerical experiments show that the algorithm is robust against common signal manipulations and attacks, such as lossy compression, re-sampling, low pass filter, noise, and so on.更多还原