Hiding image into image with hybrid attention mechanism based on GANs

• DOI: 10.1049/ipr2.13127
• Publication Date: 2024-06-18T05:40:15Z
• Authors: Yuling Zhu; Yunyun Dong; Bingbing Song; Shaowen Yao
• Keywords: Robustness; Steganography tools; Steganalysis

Abstract Image steganography is the art of concealing secret information within images to prevent detection. In deep‐learning‐based image steganography, a common practice fuse with cover directly generate stego image. However, not all features are equally critical for data hiding, and some insignificant ones may lead appearance residual artifacts in this article, novel network architecture hybrid attention mechanism based on generative adversarial introduced. This model consists three subnetworks: generator images, an extractor extracting discriminator simulate detection process, which aids producing more realistic images. A specific (HAM) module designed that effectively fuses across channel spatial domains, facilitating adaptive feature refinement deep representations. The experimental results suggest HAM only enhances quality during both extraction processes but also improves model's undetectability. Stego mixed varying levels noise training can further improve robustness. Finally, it verified outperforms current approaches datasets exhibits good