Novel Coverless Steganography Method Based on Image Selection and StarGAN

With the development of steganalysis technology keeps progressing, the traditional image steganography has confronted some security bottleneck, as the secret embedding method destroys the naturality of cover image. To overcome this difficulty, cover less image steganography (CIS) has been established for selecting images which contains secrets, and it has received widespread attention due to its merits of persevering the cover image which contains secret information. However, the existing CIS methods have the limitation of being restricted by the size of image database. Up to now, the maximum hidden capacity by CIS is only 18 bits. In order to breakthrough this limitation, this paper proposes a new CIS method based on image selection and Star Generative Adversarial Network (StarGAN). In our scheme, a traditional CIS method is used to select a natural image from database to represent the first part of the secret information, and then an image mapping is established between the rest of the secret information and the face attributes. Finally, the StarGAN is used to construct a high-quality stego image with the mapping relationship. Compared with the existing CIS methods, our method can increase the hidden capacity and maintain a better image quality, while being stronger in terms of robustness and security performance.