Open Access

ARTICLE

Reversible Data Hiding Algorithm in Encrypted Images Based on Adaptive Median Edge Detection and Ciphertext-Policy Attribute-Based Encryption

by Zongbao Jiang, Minqing Zhang^*, Weina Dong, Chao Jiang, Fuqiang Di

Key Laboratory of Network and Information Security of People’s Armed Police, Chinese People’s Armed Police Force Engineering University, Xi’an, 710086, China

* Corresponding Author: Minqing Zhang. Email:

Computers, Materials & Continua 2024, 81(1), 1123-1155. https://doi.org/10.32604/cmc.2024.055120

Received 17 June 2024; Accepted 19 August 2024; Issue published 15 October 2024

Abstract

With the rapid advancement of cloud computing technology, reversible data hiding algorithms in encrypted images (RDH-EI) have developed into an important field of study concentrated on safeguarding privacy in distributed cloud environments. However, existing algorithms often suffer from low embedding capacities and are inadequate for complex data access scenarios. To address these challenges, this paper proposes a novel reversible data hiding algorithm in encrypted images based on adaptive median edge detection (AMED) and ciphertext-policy attribute-based encryption (CP-ABE). This proposed algorithm enhances the conventional median edge detection (MED) by incorporating dynamic variables to improve pixel prediction accuracy. The carrier image is subsequently reconstructed using the Huffman coding technique. Encrypted image generation is then achieved by encrypting the image based on system user attributes and data access rights, with the hierarchical embedding of the group’s secret data seamlessly integrated during the encryption process using the CP-ABE scheme. Ultimately, the encrypted image is transmitted to the data hider, enabling independent embedding of the secret data and resulting in the creation of the marked encrypted image. This approach allows only the receiver to extract the authorized group’s secret data, thereby enabling fine-grained, controlled access. Test results indicate that, in contrast to current algorithms, the method introduced here considerably improves the embedding rate while preserving lossless image recovery. Specifically, the average maximum embedding rates for the (3, 4)-threshold and (6, 6)-threshold schemes reach 5.7853 bits per pixel (bpp) and 7.7781 bpp, respectively, across the BOSSbase, BOW-2, and USD databases. Furthermore, the algorithm facilitates permission-granting and joint-decryption capabilities. Additionally, this paper conducts a comprehensive examination of the algorithm’s robustness using metrics such as image correlation, information entropy, and number of pixel change rate (NPCR), confirming its high level of security. Overall, the algorithm can be applied in a multi-user and multi-level cloud service environment to realize the secure storage of carrier images and secret data.

---

Keywords

---