Ciphertext Computation Based On Additive Secret Sharing And Its Application In Reversible Information Hidin

The frequent information leakage issues in recent years have caused continued public concern about data privacy.In this context,the privacy-preserving techniques have been developed significantly.Among them,reversible data hiding in encrypted domain,as an emerging reversible data hiding technology,can protect the privacy of cover from data hiders and other entities in the network.On the other hand,it embeds data on cover for covert communication,authority authentication,copyright protection,etc.,and can also retain the ability to recover original cover after data extraction.Researchers have made progress in terms of embedding capacity,visual quality of decrypted images,computational complexity,ciphertext expansion,and separability through different cryptographic techniques.Meanwhile in the last five years,researchers have introduced multi-party embedding models to cope with variable embedding requirements.However,the extant works require a trade-off among the above metrics due to the inadequacy of the cryptographic techniques used.Therefore,in this paper,we first propose an efficient additive secret sharing-based privacy-preserving computation scheme,and then apply it to reversible data hiding to achieve the cover privacypreserving.The main contributions of this paper contain the following two aspects.1)Based on additive secret sharing,a privacy-preserving computation scheme is proposed.The scheme allows the proper homomorphism to be achieved at the right time for computation on ciphertext by introducing multiplicative secret sharing and combining its multiplicative homomorphism with the additive homomorphism of additive secret sharing.And based on this,a series of secure multiparty computation protocols are designed for all basic elementary functions,as well as comparison.Theoretical analysis show that the designed protocols are secure under the semi-honest model and have low communication complexity.Among them,the nonlinear function protocols are defined directly over the real field,while the error performance of these protocols on finite precision floating-point implementations is experimentally proven to be stable.2)Based on the above privacy-preserving computation scheme,a universal framework is proposed for reversible data hiding in encrypted domain with multiple hiders.The scheme can convert the traditional reversible data hiding algorithms in plaintext domain to ciphertext domain without performance loss,which cannot be achieved by existing schemes.Since the framework supports most existing reversible data hiding algorithms,the data hiders can use different embedding strategies according to their practical needs.The theoretical analysis and experimental results demonstrate that the scheme can achieve full separability with case of multiple hiders,and also significantly improves embedding capacity,visual quality of marked images,etc.