Secure Content-Based Image Retrieval Scheme for Multi-User Scenario Over Cloud

Abstract

Since the public cloud is not a credible entity when providing image retrieval services through the public cloud, it may steal the sensitive information of the image data. In recent years, cipher text image retrieval methods have been proposed to protect image privacy. However, the traditional privacy protection image retrieval scheme has low search efficiency and cannot support multi-user scenarios. Therefore, a safe and efficient multi-user outsourcing image retrieval scheme based on access control is proposed. This scheme adopts one-time encryption and matrix transformation methods to realize cipher text image retrieval based on the similarity of Euclidean distance (referred to as Euclidean distance). It uses matrix decomposition and proxy re-encryption to learn multi-user outsourced image retrieval. In addition, the local sensitive hash algorithm is used to construct the index to improve the efficiency of cipher text image retrieval. In particular, a lightweight access control strategy based on role polynomial functions is proposed, which can flexibly set image access permissions and prevent malicious users from stealing private information. Security analysis demonstrates that the proposed scheme can protect the confidentiality of images and query requests; the experimental results show that the proposed method can achieve efficient image retrieval.