Privacy preserving reversible watermarking in the encrypted domain through self-blinding

Abstract

This paper presents robust image-based reversible watermarking in the encrypted domain. Two algorithms with a high embedding rate are proposed for embedding data in a homomorphic-encrypted domain using the Paillier encryption scheme. Both algorithms exploit the self-blinding property of the Paillier scheme to accomplish flexibility in extraction. Using these algorithms, blind and error-free watermark extraction is possible in the plaintext domain and encrypted domain. The robustness of the algorithms has validated by considering various noise attacks in the encrypted domain. The proposed methods outperform its predecessors on the same embedding platform either in the flexibility of the extraction process or in the embedding capacity and find applications in privacy preserving distributed signal processing which is a major requirement in a cloud environment.