Generative Network Based Image Encryption

The algorithm which we proposed in this paper is a combined approach of Image steganography and Generative Adversarial Network. In the recommended encryption algorithm, the image matrix is processed through multiple layers of encryption where it encodes the input image matrix using steganography inside a completely new random image. Then, the newly generated image is passed through the GAN (Generative Adversarial Network) layer where the input image matrix is encrypted into a completely new model. Hence, one needs to have the same trained model to encrypt and decrypt the image, which makes it a symmetric approach where the trained model STEG file will act as a key.