Introduction
Reversible Data Hiding in Encrypted Images (RDHEI) is a critical technology for secure cloud storage. It allows a data hider (e.g., a cloud provider) to embed additional information into an encrypted image without accessing the original content. The goal is to ensure that an authorized receiver can perfectly recover both the hidden data and the original image.
However, existing RDHEI methods often face a "capacity-reversibility conflict." Traditional bit-level encryption destroys the spatial correlation between neighboring pixels, making it difficult for the receiver to accurately reconstruct the original image. This often results in high error rates or severely limited embedding capacity.
Proposed Methodology
The authors propose a novel separable RDHEI scheme that leverages block-level encryption and selective embedding to overcome these limitations.
Block-Level Encryption and Scrambling
Unlike standard stream ciphers, the proposed method uses block-level encryption. By encrypting all pixels within a small block using the same bit-sequence, the relative differences between pixels are preserved. This maintains the local spatial correlation (smoothness) within the encrypted domain, which is essential for error-free data extraction. To ensure security, a block-level scrambling step is added to obscure the image's global structure.
Selective Embedding and Block Classification
The system classifies image blocks into two categories: Embeddable Blocks (EB) and Non-Embeddable Blocks (NEB). Data is only hidden in "reliable" blocks where the algorithm guarantees 100% accurate extraction. A "position file" is generated to track these blocks and is itself hidden within the image using traditional reversible techniques.
Separability
A key feature of this method is its "separability." As shown in the framework, the receiver can perform three independent actions based on the keys they possess:
- Data Extraction Only: Using the data-hiding key, the receiver can extract the hidden bits without decrypting the image.
- Image Decryption Only: Using the decryption key, the receiver can obtain a high-quality version of the image.
- Full Recovery: Using both keys, the receiver can perfectly recover both the hidden data and the original image.
Experimental Results
The method was evaluated using standard benchmark images from the USC-SIPI database, such as "Lena" and "Baboon." The results demonstrate a significant performance leap over state-of-the-art methods:
- High Capacity: For the "Lena" image, the proposed method achieved an embedding capacity of 31,405 bits—nearly 30 times higher than traditional methods like Zhang’s (1,764 bits).
- Zero Error Rate: By selectively embedding data only in reliable blocks, the method maintains a 0% error rate, even at high capacities where other methods see error rates up to 37%.
- Visual Quality: The directly decrypted images maintain high visual fidelity, ensuring the viewing experience is not compromised by the hidden data.
Conclusion
By shifting from bit-level to block-level encryption and implementing a rigorous block selection process, this paper provides a robust solution to the capacity-reversibility conflict in RDHEI. The proposed method offers a massive increase in storage capacity while guaranteeing perfect data extraction and image recovery, making it highly suitable for secure cloud-based image management.
Summary updated