Hybrid CBIR Framework with Lattice-Based Encryption and SMPC for Post-Quantum Cloud Security

Abstract

The evolving landscape of quantum computing presents significant challenges to the security of cloud-based image retrieval systems, necessitating robust frameworks for safeguarding sensitive visual data. This paper introduces a hybrid Content-Based Image Retrieval (H-CBIR) architecture that integrates advanced post-quantum cryptographic techniques to enhance data privacy in cloud environments. To enable privacy-preserving query processing and result aggregation, the suggested system integrates secure multiparty computation (SMPC) with lattice-based encryption, which is renowned for its quantum resistance. In this design, SMPC protocols allow cooperative processing across dispersed servers without disclosing raw data, while image attributes are encrypted using lattice methods to mitigate potential quantum assaults. Experimental validation demonstrates the framework's efficacy in maintaining high retrieval accuracy alongside rigorous privacy guarantees, even under adversarial quantum-inspired threat models. The experimental validation demonstrates the proposed hybrid framework is capable of returning a similar (> 94.5%) retrieval accuracy, whilst providing a strong guarantee of privacy (leakage < 0.1%) in the quantum-inspired modeled adversarial setting. The findings serve to demonstrate the feasibility of pairing horizontal post-quantum encryption with CBIR systems to safeguard the ever-growing concern of cloud privacy breaches. In order to further strengthen privacy and performance in changing cloud infrastructures, future research will investigate the integration of new post-quantum approaches, such as code-based and multivariate cryptography, and expand the framework to broader multimodal data scenarios.