Reversible Image Steganography Using Deep Learning Method: A Review

In this paper, a complete study of reversible image steganography utilizing deep learning techniques is presented. Additionally, the traditional image steganography technique is briefly explored. The review focuses on evaluating various image steganography techniques based on their security, embedding capacity, and invisibility of secret content. Reversible steganography methods can be categorized into three major categories: Traditional, hybrid, and fully deep learning. The paper provides a clear view of significant steganography techniques with their benefits and drawbacks. The performance of different steganography methods is evaluated using the results that researchers have reported on benchmark datasets such as ImageNet, and USC-SIPI (the dataset link is provided prior to the references section). The data, when analyzed, shows that there is room for the development of new relevant deep learning architectures, which have the ability to increase the capacity of image steganography as well as its invisibility.