Verena Lachner, Rainer Böhme
Video is an ideal medium for covert communication due to its traffic volume and the many possible embedding domains offered by motion-compensated compression. This review of 30 years of video steganography research, including steganalysis and deep learning approaches, proposes a unified taxonomy and uses it to map developments across 347 sources. Many of these sources simply apply image steganography to video, and only a few critically evaluate security. The review identifies connections between these papers and highlights common pitfalls that hinder comparability. The state-of-the-art is characterized by a discussion of selected landmark papers in each embedding domain, followed by recommendations for future work.