Research map / Image generation / Autoregressive and latent generation
Towards Provably Secure and Highly Robust Generative Image Steganography Leveraging Latent Diffusion Model
Chengsheng Yuan, Zhaonan Ji, Qi Cui, Zhili Zhou, Xinting Li, Zhihua Xia
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| raise embedding threshold p from 0.4 to 1.0 (skip small-magnitude latents) (ablation) | ADPM steganography · Flicker8K PNG | extraction accuracy (%): 97.68 -> 99.84 | Improved |
| raise threshold p from 0.4 to 1.0 (ablation) | ADPM steganography · Flicker8K PNG | hiding capacity (bits): 11384 -> 5217 | Hurt |
| raise threshold p from 0.4 to 1.0 (ablation) | ADPM steganography · Flicker8K JPEG 50 | extraction accuracy (%): 93.19 -> 98.51 | Improved |
| raise threshold p from 0.4 to 1.0 (ablation) | ADPM steganography · Flicker8K median blur 7 | extraction accuracy (%): 89.70 -> 97.57 | Improved |
| sign-based distribution-preserving mapping vs RGSD-Stego under Gaussian blur 7 | ADPM steganography · Flicker8K | extraction accuracy (%): 84.95 -> 97.74 | Improved |
| sign-based mapping vs LDStega under JPEG 50 | ADPM steganography · Flicker8K | extraction accuracy (%): 94.41 -> 96.74 | Improved |
| sign-based mapping vs RGSD-Stego under median blur 7 | ADPM steganography · Flicker8K | extraction accuracy (%): 95.11 -> 93.82 | Mixed |
| distribution-preserving mapping | ADPM steganography · Flicker8K steganalysis | detection error Pe: 0.499 / 0.501 / 0.498 | No clear effect |
| ours vs RGSD-Stego | ADPM steganography · Capacity | hiding capacity (bits): 16384 -> 11384 | Hurt |
Closest papers by meaning
- SSHR: More Secure Generative Steganography with High-Quality Revealed Secret Images (ICML 2025 · similarity 0.91)
- Provably Secure Image Robust Steganography via Cross-modal Error Correction (AAAI 2025 · similarity 0.90)
- MDDM: Practical Message-Driven Generative Image Steganography Based on Diffusion Models (ICML 2025 · similarity 0.89)
- StegaStyleGAN: Towards Generic and Practical Generative Image Steganography (AAAI 2024 · similarity 0.88)
- Hiding Images in Diffusion Models by Editing Learned Score Functions (CVPR 2025 · similarity 0.87)
- ROBIN: Robust and Invisible Watermarks for Diffusion Models with Adversarial Optimization (NeurIPS 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
