Toward High Capacity and Robust JPEG Steganography Based on Adversarial Training

<div>The architecture of the HRJS. The encoder is used to generate the DCT coefficients of the stego image, while the decoder is used to reconstruct the original secret message. The attack module simulates JPEG compression. The IDCT module and DCT module convert the image into spatial domain and JPEG domain, respectively. The discriminator is used to identify whether a given image conceals the secret message.</div>

Security and Communication Networks

fig2

Figure 2

Figure 2: Toward High Capacity and Robust JPEG Steganography Based on Adversarial Training 