DTAMS: High-Capacity Generative Steganography via Dynamic Multi-Timestep Selection and Adaptive Deviation Mapping in Latent Diffusion

TL;DR

This paper introduces DTAMS, a novel generative steganography framework that significantly increases embedding rates and robustness by dynamically selecting diffusion steps and employing adaptive mapping strategies.

Contribution

The paper presents a dynamic multi-timestep selection mechanism and adaptive deviation mapping in latent diffusion models, enhancing embedding capacity and security in steganography.

Findings

Achieves 12 bpp embedding rate with high security.

Reduces average extraction error rate by 59.39%.

Outperforms state-of-the-art methods across multiple metrics.

Abstract

With the rapid development of AIGC technologies, generative image steganography has attracted increasing attention due to its high imperceptibility and flexibility. However, existing generative steganography methods often maintain acceptable security and robustness only at relatively low embedding rates, severely limiting the practical applicability of steganographic systems. To address this issue, we propose a novel DTAMS framework that achieves high embedding rates while ensuring strong robustness and security. Specifically, a dynamic multi-timestep adaptive embedding mechanism is constructed based on transition-cost modeling in diffusion models, enabling automatic selection of optimal embedding timesteps to improve embedding rates while preserving overall performance. Meanwhile, we propose a global sub-interval mapping strategy that jointly considers mapping errors and the frequency distribution of secret information, converting point-wise perturbations into interval-level statistical mappings to suppress error accumulation and distribution drift during multi-step diffusion processes. Furthermore, a multi-dimensional joint constraint mechanism is introduced to mitigate distortions caused by repeated latent-pixel transformations by jointly regularizing embedding errors at the pixel, latent, and semantic levels. Experiments demonstrate that the proposed method achieves an embedding rate of 12 bpp while maintaining excellent security and robustness. Across all evaluated conditions, DTAMS reduces the average extraction error rate by 59.39%, representing a significant improvement over SOTA methods.