Information Security Using Image Steganography Based on Generative Adversarial Networks

Abstract

Nowadays in the field of covert communication image steganographic algorithms had become one of the most common methodology and these algorithms are used to hide the secret information such as audio, video and image files in rich textured regions to decrease the chance of being identified. In Recent times i.e., from the year 2014 the generative adversarial networks i.e., GAN's had become the most well-known architectures in the area of image steganography. However, this Generative Adversarial based image steganographic algorithms will directly hide the secret information on the entire images taken, that means here the secret information is hidden on both foreground and background part of an image, so by doing this it will not consider the regional texture complexity into account, and also this may lead to compromise with the antidetection capability also. So, to solve this problem, this research work proposes a new image steganographic technique on the generated foreground region of an image with more textures. More specifically, it can be said that by using this generative adversarial method, the foreground region of an image is generated on the given cover image at the exact position and also the secret information is embedded into the foreground region of an image at the same time during the generation of a region. In this paper, different methods of generative adversarial networks are studied and discussed.