A High-Capacity Reversible Data Hiding for Encrypted JPEG Images Based on Multi-Domain Embedding

Reversible data hiding in encrypted JPEG images (JPEG-RDH-EI) is one of the key technologies for securely storing and managing confidential images in cloud environments. However, existing methods typically operate only in the coefficient domain or the encoding domain, failing to fully exploit the characteristics of encrypted images, which results in limited embedding capacity and significant file expansion. To address these limitations, this paper proposes a high-capacity JPEG-RDH-EI scheme based on multi-domain embedding. To reduce file expansion, a rotation model is devised that leverages the correlation of DCT coefficients within a range that does not notably alter the run length. This model embeds secret data by strategically rotating and rearranging these DCT coefficients. On this basis, the scheme constructs an optimal VLC mapping relationship and adaptively adjusts embedding parameters according to the characteristics of the encrypted image, based on the RSV histogram shifting technique, achieving data embedding in the encoding domain, thereby further enhancing embedding capacity. The proposed scheme supports fully separable data extraction and image recovery, making it suitable for various application scenarios. Evaluation results demonstrate that the maximum embedding capacity of this scheme is 2 to 3 times greater than that of existing advanced schemes, with some images achieving a maximum embedding capacity of over 100,000 bits. Moreover, the proposed scheme achieves an average unit file expansion of only about 0.03, outperforming current advanced methods.