Answers are grounded in this paper and cite their sources.
Computational framework demonstrates accurate visual security evaluation in perceptual image encryption, indicating enhanced detection of localized information leakage.
Visual security quantification is critical for evaluating perceptual image encryption schemes. Traditional visual security indices often rely on partial feature extraction and fixed-position block matching, which limits their capability to capture localized information leakage. To address these limitations, this paper proposes a novel Multi-Feature Fusion Visual Security Index (MFFVSI). Within this framework, an optimal block-matching strategy is designed to calculate local feature similarity (LFS), enabling the detection of displaced sensitive information leakage. Concomitantly, global feature similarity (GFS) is derived in the Gaussian-weighted frequency domain to compensate for structural distortion induced by block partitioning. LFS and GFS are subsequently integrated via a multi-resolution regression module to output the final visual security score. Experimental results demonstrate that the proposed method exhibits strong consistency with subjective visual perception and high universality across various perceptual encryption schemes, making it suitable for practical visual security evaluation applications.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: