Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-User Multi-Key Image Steganography with Key Isolation

Tzu-Ti Wei, Yu-Han Tseng, Jun-Yi Lin, Yu-Chee Tseng, Jen-Jee Chen

Abstract

Steganography conceals secret information within innocuous carriers while preserving visual fidelity and enabling reliable recovery. Recent unified networks operate normally under untriggered conditions but switch to hidden steganographic tasks when triggered. PUSNet follows this paradigm by performing image purification during normal operation and steganographic embedding when activated. However, it supports only a single user with one key pair, limiting its applicability in multi-user settings. We propose PUSNet-MK, a multi-key extension that enforces strict key isolation via a mismatched-key isolation loss, effectively preventing cross-key decoding when a wrong key is applied. This design preserves the intended steganographic behavior while addressing a critical security limitation of PUSNet. Extensive experiments demonstrate that PUSNet-MK produces high-quality stego images and accurate secret recovery, while preventing unintended information leakage.

Multi-User Multi-Key Image Steganography with Key Isolation

Abstract

Steganography conceals secret information within innocuous carriers while preserving visual fidelity and enabling reliable recovery. Recent unified networks operate normally under untriggered conditions but switch to hidden steganographic tasks when triggered. PUSNet follows this paradigm by performing image purification during normal operation and steganographic embedding when activated. However, it supports only a single user with one key pair, limiting its applicability in multi-user settings. We propose PUSNet-MK, a multi-key extension that enforces strict key isolation via a mismatched-key isolation loss, effectively preventing cross-key decoding when a wrong key is applied. This design preserves the intended steganographic behavior while addressing a critical security limitation of PUSNet. Extensive experiments demonstrate that PUSNet-MK produces high-quality stego images and accurate secret recovery, while preventing unintended information leakage.
Paper Structure (10 sections, 7 equations, 5 figures, 1 table)

This paper contains 10 sections, 7 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Deep Learning-based Steganography
  4. Multi-user and Multi-key Learning
  5. Proposed PUSNet-MK
  6. Review of PUSNet
  7. Multi-key Extension and Training Objectives
  8. Training Objective and Key Separation
  9. Experiments
  10. Conclusions

Figures (5)

  • Figure 1: PCA. Recovery features in PUSNet using matched and mismatched keys reveal feature entanglement. PUSNet-MK resolves this problem through key isolation.
  • Figure 2: PUSNet-MK framework. Green blocks are shared and learnable, while others are frozen and key-conditioned.
  • Figure 3: Qualitative comparison. Stego and recovered images, with residuals shown in the lower-right corner.
  • Figure 4: Cross-key decoding tests.$k^j_r (I_{\text{stego}}^{i}) = N[W \odot M + W_r^{j} \odot \overline{M}](I_{\text{stego}}^{i})$ denotes decoding $I_{\text{stego}}^{i}$ with a mismatched key $k^j_r$, $i \neq j$, where $rand$ means a random key.
  • Figure 5: Scalability study. We vary the number of keys ($K$) and sparse ratio ($\alpha$). Each confusion matrix shows recoverability, where rows correspond to stego images encoded with different keys and columns correspond to keys used for decoding. In each cell, we report PSNR (black) and SSIM (brown).