Image steganography based on generative implicit neural representation

TL;DR

This work addresses the limits of resolution and extractor training in image steganography by introducing GINR-Stega, which represents cover images as continuous functions via a function generator trained with a generative adversarial stochastic process and uses a fixed neural network as the extractor to decouple training from the extractor. The approach achieves resolution independence and substantially lowers extractor training costs, enabling fast embedding and exact message recovery (e.g., 100% accuracy at 3 bpp for 64×64 images) while maintaining perceptual image quality. Key contributions include the first integration of a function generator with steganography, grid-independent data representation through INR, and a fixed-extractor design that transfers training to the image, with empirical results on CelebA and ERA5 datasets showing competitive image quality, efficiency, and modest Stegexpose detectability. The method broadens multimedia carrier compatibility, reduces training overhead, and offers a scalable pathway for high-capacity, covert communication in diverse data modalities.

Abstract

In the realm of advanced steganography, the scale of the model typically correlates directly with the resolution of the fundamental grid, necessitating the training of a distinct neural network for message extraction. This paper proposes an image steganography based on generative implicit neural representation. This approach transcends the constraints of image resolution by portraying data as continuous functional expressions. Notably, this method permits the utilization of a diverse array of multimedia data as cover images, thereby broadening the spectrum of potential carriers. Additionally, by fixing a neural network as the message extractor, we effectively redirect the training burden to the image itself, resulting in both a reduction in computational overhead and an enhancement in steganographic speed. This approach also circumvents potential transmission challenges associated with the message extractor. Experimental findings reveal that this methodology achieves a commendable optimization efficiency, achieving a completion time of just 3 seconds for 64x64 dimensional images, while concealing only 1 bpp of information. Furthermore, the accuracy of message extraction attains an impressive mark of 100%.

Figures (15)

• Figure 1: The basic idea of our scheme.
• Figure 2: The Framework of GINR-Stega.
• Figure 3: The Construction of Function Generator.
• Figure 4: Training of Function Generator.
• Figure 5: Superresolution.