FSAIR: Fine-Grained Secure Approximate Image Retrieval for Mobile Cloud Computing

Cloud computing and the Internet of Things (IoT) provide robust technological support for the development of image retrieval services. Specifically, images are highly sensitive and private data in e-healthcare and security surveillance. Existing retrieval schemes often do not strike a good balance between privacy protection and retrieval performance. It makes data vulnerable to illegal attacks and leads to high retrieval costs, thereby affecting the overall quality and usability of the system. To address these issues, this paper introduces a fine-grained secure approximate image retrieval (FSAIR) scheme for mobile cloud computing. Our approach implements a multi-verification architecture that provides precise identity control and an untraceable strategy. Furthermore, FSAIR constructs a flexible and secure hierarchical structure to support the efficient retrieval of large-scale high-dimensional data. By introducing Bloom filters to replace index nodes, FSAIR ensures system efficiency and employs bit-pattern encoding descriptors to search approximate data without violating privacy. We show that the proposed scheme protects data under different threat modes through security analysis. We also demonstrate the feasibility and superiority of the proposed solution through experimental evaluation, using the Caltech-256 dataset.