Information Technology Reference
In-Depth Information
subset of the image pixels and/or a bit-wise approach are described in [3], [4], [6],
[16], [20], [22], [28], [29], etc. Transform-based techniques, that operate images rep-
resented by a finite set of orthogonal or bi-orthogonal “basis functions” are presented
in [5], [10], [15], [19], [23], [24], [26], etc. Examples of such transforms are Discrete
Cosine, Wavelet transform, Singular Value Decomposition, etc. A specific technique
is provided by fractal-based approach that constructs a “fractal code” encoding both
the cover image and hidden message [25].
It should be noticed that most developed techniques are watermarking-oriented
which major requirement is robustness to a wide range of distortions, but not of the
high rate of the embedded data. However, many military and industrial applications
like hidden communication (HC), in particular, hidden transmission of digital images
call for both invisibility of high volumes of data and survivability of the transmitted
information. Examples are transmission of industrial secrets, plans of covert opera-
tions [12], etc. Unfortunately, most existing techniques capable of providing the ratio
of transparently embedded data, required for HC, are highly sensitive to many distor-
tions, in particular, to lossy compressions like JPEG .
In contrast, paper [19] presents a technique oriented to HC application. It proposes
an approach called Spread Spectrum Image Steganography). It is a blind scheme in
which one does not need to use the cover image in order to extract the hidden mes-
sage. The central point of this approach is to embed the message in the form of a sig-
nal provided that the signal has the same characteristics as noise inherent to this im-
age. The last property is provided by a special modulation of the message data.
HC application is the focus of paper [10], which presents a technique primarily
destined for meeting the requirements of the covert communication. The technique
uses Singular Value Decomposition (SVD) transform of digital images. According to
the properties of SVD of a digital image, each singular value (SV) specifies the lumi-
nance (energy) of an SVD image layer, whereas the respective pair of singular vectors
specifies the image geometry. Accordingly, slight variations of SVs cannot affect the
visual perception of the quality variations of the cover image. The robustness of the
approach is provided by the fact that it embeds of data through slight modifications of
the largest SVs of a small block of the segmented covers, i.e. it embeds a message
into low frequency of a cover image. Simulation has proved this anticipation.
This paper presents the results of the detailed simulation-based study of the HC
critical properties of SVD-based data hiding. Primarily, the paper studies interde-
pendencies of both the capacity of invisibly embedded data and robustness of its
transmission through communication channel distorted by JPEG compression, on the
one hand, and values of attributes determining the embedding procedure, on the other
hand. The simulated situation assumes that a message is embedded into digital image
(cover) represented in the bmp format and the resulting image is transmitted in the
JPEG format. The received image with embedded data is again transformed to the
JPEG format. Thus, the cover image can be considered as a noisy channel distorted
by the JPEG compression. In addition, it is assumed that the channel can also be in-
tentionally distorted by applying of a JPEG compression to the transmitted cover with
embedded data. As the robustness metric the value of Bit Error Rate (BER) is used.
The rest of the paper is organized as follows. In Section 2 the concept of the SVD-
based approach to and algorithm for hiding data in digital images is briefly explained
and one of the previously developed algorithms is outlined. Section 3 describes a
distortion model as applied to a message transmitted through a steganographic chan-
Search WWH ::




Custom Search