Аннотация:
Steganography currently remains one of the most rapidly and effectively developing areas of
information security. It leads to an increase in the relevance of steganalysis to avoid the
negative consequences of covert communication for the purpose of illegal actions. Preference
in the organization of a covert communication channel today is given to digital video.
However, modern video steganalysis is behind the required level, particularly in conditions of
low capacity of the covert channel. This paper is dedicated to developing a new effective
steganalysis method for detecting embedded additional information in digital video (a sequence
of digital images) by one of the most common steganographic methods today, which is the
LSB method. The developed method is based on the sensitivity of linearity of singular vectors
frequency in the corresponding original content matrices to disturbing influences. The authors
determined this property in previous works. The current work established qualitative and
quantitative differences between the properties of function, which represents the dependency
of the singular vector frequency in the image/video frame matrix on its number for original
contents and steganographic messages, which made it possible to develop an algorithmic
implementation of the proposed method. The efficiency of this implementation in the case of
a low capacity of a covert communication channel (<0.25 bit/pixel), including at a low frame
embedding rate (<50%), exceeds the efficiency of analogues, which made it possible to
increase the efficiency of steganalysis of digital video (sequence of digital images) as a whole.
