In recent years, the increasing requirements in cloud storage and cloud computing have made it necessary to encrypt digital images for privacy protection. Meanwhile, many reversible data hiding (RDH) algorithms in the encrypted domain have been proposed. However, most of these algorithms are for gray-level images, and the intrinsic cross-channel correlations of color images cannot be utilized to improve the embedding capacity. In this paper, we propose a novel data hiding method for encrypted color images. In the encryption stage, the homomorphic property of encryption is achieved by basic modular addition. During the data hiding process, the cross-channel correlations between R, G and B channels are generated in encrypted domain, and data hiding is performed by the difference histogram shifting. Analysis and experiments demonstrate that the proposed method is secure and the RDH performance is superior.

近年来，对云存储和云计算的需求不断增加，使得对数字图像进行加密以保护隐私成为必要。同时，已经提出了许多加密域中的可逆数据隐藏（RDH）算法。然而，这些算法大多是针对灰度图像的，不能利用彩色图像的内在跨通道相关性来提高嵌入能力。在本文中，我们提出了一种新的加密彩色图像数据隐藏方法。在加密阶段，加密的同态性是通过基本的模加法来实现的。在数据隐藏过程中，R、G、B通道之间的跨通道相关性在加密域中产生，通过差异直方图移位进行数据隐藏。