Training-Free Coverless Multi-Image Steganography with Access Control
Researchers have developed MIDAS, a novel training-free diffusion-based framework for coverless image steganography. This system allows for hiding information across multiple images with user-specific access controls, a critical feature for privacy-sensitive applications. MIDAS utilizes a Random Basis mechanism and a Latent Vector Fusion module to enhance image quality, diversity, and resistance to steganalysis, outperforming existing methods in its access control capabilities. AI
