The secure judgment of graphic similarity against malicious adversaries
Xin Liu, Yang Xu, Gang Xu, Xiu‐Bo Chen · Research Square · 2022
Abstract With the the advent era of big data, the secure computation calculates data on the premise of protecting data privacy, to realize the availability and invisibility of data. Secure multi-party computation, as one of three major technical tools of privacy computing, can still securely carry out data collaborative computation without a trusted third party. As an important branch of secure multi-party computation, the secure computing geometric problem can solve practical problems in the military, national defense, finance, life, and other fields, which has important research significance. In this paper, the graphic similarity problem is studied. Firstly, this paper proposes the adjacency matrix vector coding method of isomorphic graphics and uses the Paillier variant cryptosystem to securely solve the graphic similarity judgment under the semi-honest model. By using an elliptic curve cryptosystem and zero-knowledge proof to solve the possible malicious attacks under the semi-honest model, a graphic similarity judgment protocol under the malicious model is designed. The protocol can resist malicious attacks, has high computational efficiency, and has wide application value.