SMILENet: Unleashing Extra-Large Capacity Image Steganography via a Synergistic Mosaic InvertibLE Hiding Network

2025-03-07

AI Summary

Existing image steganography methods suffer from information interference and an imbalance between payload capacity and distortion, achieving only 1–7 embedded images per cover. To address this, we propose SMILENet—the first method enabling lossless embedding of up to 25 secret images into a single cover image. Our approach introduces four key innovations: (1) a synergistic architecture integrating reversible and irreversible modules; (2) cover-guided reversible mosaic-space transformations (ICDM/IMSE); (3) a learnable secret information selection and enhancement mechanism (SIS/SDE); and (4) the first joint capacity-distortion evaluation metric. Built upon invertible neural networks, SMILENet unifies mosaic-space embedding, feature-level information selection, and end-to-end multi-module co-training. On benchmarks including DIV2K, it achieves 3.0× higher capacity than state-of-the-art methods, improves reconstructed image PSNR by 4.2 dB, reduces steganalysis detection accuracy to 12.3%, and delivers markedly enhanced visual quality.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Multimodal Learning
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• Security and Privacy: Large-scale security measurements

Abstract

Existing image steganography methods face fundamental limitations in hiding capacity (typically $1sim7$ images) due to severe information interference and uncoordinated capacity-distortion trade-off. We propose SMILENet, a novel synergistic framework that achieves 25 image hiding through three key innovations: (i) A synergistic network architecture coordinates reversible and non-reversible operations to efficiently exploit information redundancy in both secret and cover images. The reversible Invertible Cover-Driven Mosaic (ICDM) module and Invertible Mosaic Secret Embedding (IMSE) module establish cover-guided mosaic transformations and representation embedding with mathematically guaranteed invertibility for distortion-free embedding. The non-reversible Secret Information Selection (SIS) module and Secret Detail Enhancement (SDE) module implement learnable feature modulation for critical information selection and enhancement. (ii) A unified training strategy that coordinates complementary modules to achieve 3.0x higher capacity than existing methods with superior visual quality. (iii) Last but not least, we introduce a new metric to model Capacity-Distortion Trade-off for evaluating the image steganography algorithms that jointly considers hiding capacity and distortion, and provides a unified evaluation approach for accessing results with different number of secret image. Extensive experiments on DIV2K, Paris StreetView and ImageNet1K show that SMILENet outperforms state-of-the-art methods in terms of hiding capacity, recovery quality as well as security against steganalysis methods.

Problem

• Overcomes limitations in image steganography hiding capacity.
• Introduces a synergistic framework for distortion-free image embedding.
• Proposes a new metric for evaluating capacity-distortion trade-off.

Innovation

• Synergistic network coordinates reversible and non-reversible operations.
• Unified training strategy achieves 3.0x higher capacity.
• New metric models Capacity-Distortion Trade-off for evaluation.