A survey of steganalysis methods in the papers of foreign authors
D E Vilkhovskiy · Matematičeskie struktury i modelirovanie · 2020
The paper provides a survey of foreign studies regarding steganalysis, aimed to detect hidden message insertion made by applying least significant bit replacement and discrete cosine transform algorithms. It is noted the further steganalysis methodology development splits in two directions: a decrease of the complexity and cost of processing and detection while maintaining a high level of classification rate, which is quite justified in the case of the presence of insertions with a large payload, i.e. up to 100%; or an increase of the insert recognition efficiency when dealing with images of a low payload. Besides, during the last five years, steganalisys methods based on machine learning and deep learning began to play a dominant role in steganalysis