Generative Steganography by Sampling

Abstract

In this paper, a novel data-driven information hiding scheme called "generative steganography by sampling" (GSS) is proposed. Unlike in traditional modification-based steganography, in our method the stego image is directly sampled by a powerful generator: no explicit cover is used. Both parties share a secret key used for message embedding and extraction. The Jensen-Shannon divergence is introduced as a new criterion for evaluating the security of generative steganography. Based on these principles, we propose a simple practical generative steganography method that uses semantic image inpainting. The message is written in advance to an uncorrupted region that needs to be retained in the corrupted image. Then, the corrupted image with the secret message is fed into a Generator trained by a generative adversarial network (GAN) for semantic completion. Message loss and prior loss terms are proposed for penalizing message extraction error and unrealistic stego image. In our design, we first train a generator whose training target is the generation of new data samples from the same distribution as that of existing training data. Next, for the trained generator, backpropagation to the message and prior loss are introduced to optimize the coding of the input noise data for the generator. The presented experiments demonstrate the potential of the proposed framework based on both qualitative and quantitative evaluations of the generated stego images. INDEX TERMS Generative adversarial network, Image inpainting, Steganography

Camouflage Generative Adversarial Network: Coverless Full-image-to-image Hiding

“…To ensure a fair comparison, we use a several image hiding algorithms which have full-image-to-image hidden capacity, including [12] and [13] as benchmark methods, because both image quality and hiding undetectability are directly affected by the size of the hidden message. Further, we compare the hidden capacity of Cam-GAN with [8], [10], [11] to demonstrate superiority compared with state-of-the-art coverless image hiding algorithms.…”

“…Finally, we further compare our Cam-GAN and three coverless image hiding algorithms via image synthesis [8], [10], [11] in terms of hidden capacity as shown in Table II. From there, we can find that Cam-GAN enlarges the hidden capacity significantly compared with these state-of-the-art coverless image hiding algorithms and only our approach yields a capacity 24 bits per pixel (bpp) to enable full-image-to-image hiding.…”

Learning to Generate Steganographic Cover for Audio Steganography Using GAN

“…Unlike SGAN, Hayes et al [15] proposed to use the secret message and cover image to generate stego image. Zhang et al [16] proposed a novel data-driven information hiding scheme called "generative steganography by sampling" (GSS) that the stego image was directly sampled by a generator without using covers. Tang et al [17] combined GAN with adaptive steganography to propose automatic steganographic distortion learning framework (ASDL-GAN), in which the GAN component was supposed to learn the embedding change probability map.…”

SteganoCNN: Image Steganography with Generalization Ability Based on Convolutional Neural Network

“…A significant difference of HidingGAN is the addition of a steganalyzer for adversarial training, similar to the classic prisoner model problem. Zhou Zhang et al [ 17 ] proposed a steganography scheme based on GAN and Cardan grid mask. First, the noise vector is used as the input of GAN for training, and the output is a natural image.…”