Multi-Image Texture Feature-Guided Adaptive Steganography

Abstract

To ensure secure communication, steganographic techniques must be employed to achieve high imperceptibility, robustness, and payload capacity. The current approaches either employ texture-based adaptivity or multi-image embedding in isolation, which can lead to limited efficiency and a risk of steganalysis vulnerability. The authors propose a fresh hybrid image steganography approach that merges texture feature analysis with multi-image adaptive embedding, enhanced by adaptive alpha blending, to overcome these issues. By using texture descriptors like LBP and GLCM, the payload is translated into local and global patterns, which can guide adaptive bit placement, while being distributed across multiple cover images to ensure maximum security and capacity. By using adaptive alpha blending, the embedding strength can be adjusted dynamically, further improving imperceptibility. Despite being less robust than current schemes, the proposed method is shown to have superior performance in experiments, with higher PSNR and enhanced SSIM capabilities and embedding capacity. However, it fails to significantly reduce compression and noise attacks. The latest innovation involves the use of texture-based adaptivity, multi-image embedding, and alpha blending to create a unified framework that is both sustainable and secure for transmitting concealed data in modern digital communication methods.