Delving into the methods of coverless image steganography

Conventional cover-based image steganography methods embed secret information by modifying the original state of a cover image. This type of algorithm leaves a trace of changes on output stego image and eventually leads to successful detection by common steganalysis tools. As a solution, a coverless image steganographic method is proposed, where no cover image is required for embedding secret information. In this paper, the conventional coverless image steganography methods are first reviewed and categorized into constructive and nonconstructive-based methods. Next, these methods are summarized and analyzed, followed by a discussion about their advantages and drawbacks. Finally, the performances of the proposed methods are discussed using the common steganography evaluation metrics, including resistance to attack, embedding capacity, and perceptual image quality.