Reversible and recoverable authentication method for demosaiced images using adaptive coding technique. (December 2020)

• Record Type: Journal Article
• Authors: Zhou, Xiaoyu; Hong, Wien; Weng, Shaowei; Chen, Tung-Shou; Chen, Jeanne
• Is Part Of: Journal of information security and applications. Volume 55(2020)
• Publication Date: 2020-12
• Subjects: Authentication -- Image recovery -- Image demosaicing -- Reversibility; Computer security -- Periodicals; Information technology -- Security measures -- Periodicals; 005.805
• DOI: 10.1016/j.jisa.2020.102629
• Languages: English
• ISSNs: 2214-2126

Abstract

Abstract: This paper proposes a reversible authentication scheme for demosaiced images with the capability to approximately recover original contents of tampered parts. The existing methods protect the marked images to a large extent; however, they cannot detect some intentional modifications, or lack the self-recovery capability of tampered regions. The proposed method uses the most significant bits (MSBs) of sampled components to generate recovery codes, and embeds them into the least significant bits (LSBs) of rebuilt components. The MSBs of rebuilt components are adjusted to reduce the embedding distortion. The authentication codes are generated by hashing the MSBs of adjusted rebuilt components, and are embedded into their LSBs. By analyzing the extracted authentication and recovery codes, the tampered regions can be localized, and the original contents can be approximately recovered. If the marked image is untampered, the original demosaiced image can be exactly reconstructed. Experimental results demonstrate that the proposed method not only provides an excellent marked image quality but also achieves a satisfactory detection and recovery results.