Image steganography is the method of concealing information, which can be text, image, or video, under a cover image in a way that it is invisible to the human eyes. Payload capacity, security and robustness are the features of steganography. A high-quality image steganography ensures a large payload capacity, superior security, and robustness against adversarial attacks. Image steganography is considered to be reversible if the concealed information's original state can be restored with little or no loss in the pixel values and textures. The main goal of this paper is to look at some of the typical techniques and models that are used to build image steganography systems by adopting the Preview, Question, Read, Summarize (PQRS) method to explain how the payload capacity, security, and robustness of these models may be quantified for future research and recommend GAN-based deep learning methods for improving existing reversible image steganography models.

 

Keywords: Image steganography, GAN-based models, Payload capacity, PSNR, Reversibility, SSIM