密文图像中的可逆数据隐藏技术既能在密文中嵌入附加信息，又能无损恢复原始明文图像，是新兴的研究热点。本项目从三方面开展密文域可逆数据隐藏研究：1）针对密文域可逆隐藏缺乏理论支持这一问题，发展“明文失真—嵌入容量”率失真理论，推导明文允许失真下的密文域嵌入率上确界，并研究安全嵌入条件，为方法设计指明方向并提供评价准则。2）在理论研究的指导下，发展新机制与新方法来逼近嵌入率极限。开拓非均匀嵌入与渐进式恢复新机制，突破传统均匀嵌入与单一恢复的局限；研究面向分组加密的高效安全可逆隐藏，发展密码分组链接模式下的密文域分布式嵌入编码；深入探索数据嵌入与图像特征变化之间的制约关系，发展基于高维特征分类的密文JPEG可逆隐藏新机制。3）以理论与方法为基础，发展基于自嵌入的密文图像可逆数据隐藏，解决特定应用场景中的密文有损传输问题。研究成果将推动密文域可逆隐藏理论与方法的进步，在信息安全领域有广泛的应用前景。

The technique of reversible data hiding (RDH) in encrypted images, which requires not only the capability of embedding data in the encrypted image, but also the lossless recovery of the original image, is one of the research hotspots. This project will investigate the topic from three aspects. First of all, since there is no theory of the topic, we will develop the rate-distortion theory and the security requirements of RDH in encrypted domain, which is vital for designing RDH methods and can provide the benchmark for the performance assessments. Next, RDH methods aiming at high payloads with less distortion will be developed under the new theory, including the uneven-embedding coding and progressive-recovery approach for the stream-ciphered images, the distributed source coding in the encrypted domain for CBC based and block-ciphered images, and the ensemble classifier based secure RDH for encrypted JPEG images. At last, based on the theory and the methods, we will focus on a specific scenario with an emerging problem that some breaking errors may happen in the encrypted images when transmitted through a lossy channel. We propose a self-embedding based RDH in the encrypted images to resolve this problem. Achievements of the project will be significant for the development of RDH in encrypted images, and have a good application prospect in the information security field.