High-capacity reversible data hiding in encrypted medical images using adaptive pixel-modulation and HBP-RMC

Abstract

With the rapid development of telemedicine technology, electronic medical records (EMR) are facing security issues such as illegal access, leakage and malicious tampering during the transmission of electronic patient information (EPI). In view of this, a high-capacity RDHEI based on adaptive pixel modulation and high bit-plane based redundancy matrix coding (HBP-RMC) is proposed to effectively achieve patient privacy protection by embedding private data such as electronic medical records and authentication information into encrypted medical images. Specifically, the carrier image is encrypted using designed block-level disruption and adaptive pixel-modulation to preserve pixel-correlation. This method provides the basis for high embedding rates while ensuring the overall security of the image. Next, an accurate block-level pixel prediction is proposed to make full use of the high correlation to generate considerable redundancy and in balance with high prediction accuracy. Then the proposed HBP-RMC is used to maximally convert the redundant information into the embeddable space and effectively limit the size of the auxiliary information. At the decoding end, users can realize separable image restoration and information extraction according to their own keys. The maximum ERs of our scheme on the BOWS-2 and BOSSbase datasets can reach 2.9332 bpp and 3.0249 bpp, respectively. Extensive experiments prove that our algorithm outperforms some existing state-of-the-art schemes.