基于非对称扩展的可逆水印研究

The rapid development of information technology has led to rapid popularity of digital media in the computer network, its greatly facilitate the people's daily life and work, but it also caused another such as leakage and copyright protection. In order to protect copyright of digital information and restore the original information, reversible watermarking technology emerge as required, it supplements the defect of regular watermarking, and it makes great development in recent years, especially the expansion based reversible watermarking technology. The project is committed to reduce the expansion distortion which caused by the watermarking embedding, and prepare to research the asymmetry of reversible watermarking from the stage of error generation to watermarking embedding. The research contents not only include the general theoretical framework of the asymmetric expansion, such as asymmetric prediction scheme, the double complement embed strategy, the asymmetric distribution feature and algorithm evaluation, and so on; but also involve the key technologies (such as error, error of two-dimensional histogram, etc.) that the current hot technologies used in asymmetric expansion method, such as the prediction scheme of error pair, the two-dimensional asymmetric error expansion and directed integer transform, etc. The study provides new technology and method for the application of reversible watermarking, and provides theoretical basis and technical support for promoting the digital image security applications.

信息技术的快速发展促使数字媒体在计算机网络迅速普及，极大地便利了人们的日常生活和工作，但同时也导致了诸如隐私泄露、版权纠纷等问题。为保障这些信息的安全，数字水印技术应运而生。可逆水印是数字水印的一个重要研究领域，其能无损恢复原始载体的特性，弥补了常规水印的不足，近年来取得了巨大的发展，尤其是基于扩展的可逆水印技术。本课题拟对可逆水印从误差生成到水印嵌入各个阶段的非对称性进行深入研究，从而减少水印图像的扩展扭曲，提高水印图像质量。其研究内容不仅包括非对称扩展的一般理论框架，如非对称预测机制、双层补偿嵌入策略、非对称分布特征和算法评价等；而且还涉及当前热点扩展算法（如误差对扩展、二维误差直方图等）采用非对称手段嵌入水印时的关键技术，如误差对预测机制、二维非对称误差扩展和有向整数变换等。本课题为可逆水印研究提供新技术和方法，为推动数字图像安全应用提供理论基础和技术支撑。

可逆水印的研究对信息安全技术的健康发展具有重要意义。一方面其具有客观科学的评价指标，其研究能综合提升信息安全能力；另一方面其研究往往是其他学科的基础，对其他学科研究具有极大的促进作用。本项目主要研究内容如下：第一、研究了可逆水印的一般框架，包括非对称扩展的分布模型、多预测机制的选择、非对称选择函数以及还原率等，为可逆水印研究提供了理论基础；第二、研究了加密域可逆水印，包括同态加密、码分复用在多图像和多信号中的可逆实现，为加密云服务提供了技术支撑；第三、研究了可逆图像变换和图像伪装技术，包括灰度共生矩阵的特征匹配以及图像可逆隐藏等，为进行大容量可逆信息隐藏提供了新的思路。