First of all, the whole code of STABYLO can be downloaded
-\footnote{\url{http://members.femto-st.fr/raphael-couturier/stabylo/}}.
+\footnote{\url{http://http://members.femto-st.fr/jf-couchot/en/stabylo}}.
For all the experiments, the whole set of 10,000 images
of the BOSS contest~\cite{Boss10} database is taken.
In this set, each cover is a $512\times 512$
-\subsection{Image quality}\label{sub:quality}
-The visual quality of the STABYLO scheme is evaluated in this section.
-For the sake of completeness, three metrics are computed in these experiments:
-the Peak Signal to Noise Ratio (PSNR),
-the PSNR-HVS-M family~\cite{psnrhvsm11},
-and
-the weighted PSNR (wPSNR)~\cite{DBLP:conf/ih/PereiraVMMP01}.
-The first one is widely used but does not take into
-account the Human Visual System (HVS).
-The other ones have been designed to tackle this problem.
+% \subsection{Image quality}\label{sub:quality}
+% The visual quality of the STABYLO scheme is evaluated in this section.
+% For the sake of completeness, three metrics are computed in these experiments:
+% the Peak Signal to Noise Ratio (PSNR),
+% the PSNR-HVS-M family~\cite{psnrhvsm11},
+% and
+% the weighted PSNR (wPSNR)~\cite{DBLP:conf/ih/PereiraVMMP01}.
+% The first one is widely used but does not take into
+% account the Human Visual System (HVS).
+% The other ones have been designed to tackle this problem.
-If we apply them on the running example with the Adaptive and STC strategies,
-the PSNR, PSNR-HVS-M, and wPSNR values are respectively equal to
-68.39, 79.85, and 89.71 for the stego Lena when $b$ is equal to 7.
-If $b$ is 6, these values are respectively equal to
-65.43, 77.2, and 89.35.
+% If we apply them on the running example with the Adaptive and STC strategies,
+% the PSNR, PSNR-HVS-M, and wPSNR values are respectively equal to
+% 68.39, 79.85, and 89.71 for the stego Lena when $b$ is equal to 7.
+% If $b$ is 6, these values are respectively equal to
+% 65.43, 77.2, and 89.35.
-\begin{table*}
-\begin{center}
-\begin{small}
-\setlength{\tabcolsep}{3pt}
-\begin{tabular}{|c|c|c||c|c|c|c|c|c|c|c|c|c|}
-\hline
-Schemes & \multicolumn{4}{|c|}{STABYLO} & \multicolumn{2}{|c|}{HUGO}& \multicolumn{2}{|c|}{EAISLSBMR} & \multicolumn{2}{|c|}{WOW} & \multicolumn{2}{|c|}{UNIWARD}\\
-\hline
-Embedding & Fixed & \multicolumn{3}{|c|}{Adaptive (about 6.35\%)} & Fixed &Adaptive & Fixed &Adaptive & Fixed &Adaptive & Fixed &Adaptive \\
-\hline
-Rate & 10\% & + sample & +STC(7) & +STC(6) & 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%\\
-\hline
-PSNR & 61.86 & 63.48 & 66.55 & 63.7 & 64.65 & {67.08} & 60.8 & 62.9&65.9 & 68.3 & 65.8 & 69.2\\
-\hline
-PSNR-HVS-M & 72.9 & 75.39 & 78.6 & 75.5 & 76.67 & {79.6} & 71.8 & 76.0 &
-76.7 & 80.35 & 77.6 & 81.2 \\
-\hline
-wPSNR & 77.47 & 80.59 & 86.43& 86.28 & 83.03 & {88.6} & 76.7 & 83& 83.8 & 90.4 & 85.2 & 91.9\\
-\hline
-\end{tabular}
-\end{small}
-\end{center}
-\caption{Quality measures of steganography approaches\label{table:quality}}
-\end{table*}
+% \begin{table*}
+% \begin{center}
+% \begin{small}
+% \setlength{\tabcolsep}{3pt}
+% \begin{tabular}{|c|c|c||c|c|c|c|c|c|c|c|c|c|}
+% \hline
+% Schemes & \multicolumn{4}{|c|}{STABYLO} & \multicolumn{2}{|c|}{HUGO}& \multicolumn{2}{|c|}{EAISLSBMR} & \multicolumn{2}{|c|}{WOW} & \multicolumn{2}{|c|}{UNIWARD}\\
+% \hline
+% Embedding & Fixed & \multicolumn{3}{|c|}{Adaptive (about 6.35\%)} & Fixed &Adaptive & Fixed &Adaptive & Fixed &Adaptive & Fixed &Adaptive \\
+% \hline
+% Rate & 10\% & + sample & +STC(7) & +STC(6) & 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%& 10\%&$\approx$6.35\%\\
+% \hline
+% PSNR & 61.86 & 63.48 & 66.55 & 63.7 & 64.65 & {67.08} & 60.8 & 62.9&65.9 & 68.3 & 65.8 & 69.2\\
+% \hline
+% PSNR-HVS-M & 72.9 & 75.39 & 78.6 & 75.5 & 76.67 & {79.6} & 71.8 & 76.0 &
+% 76.7 & 80.35 & 77.6 & 81.2 \\
+% \hline
+% wPSNR & 77.47 & 80.59 & 86.43& 86.28 & 83.03 & {88.6} & 76.7 & 83& 83.8 & 90.4 & 85.2 & 91.9\\
+% \hline
+% \end{tabular}
+% \end{small}
+% \end{center}
+% \caption{Quality measures of steganography approaches\label{table:quality}}
+% \end{table*}
-Results are summarized in Table~\ref{table:quality}.
-In this table, STC(7) stands for embedding data in the LSB whereas
-in STC(6), data are hidden in the last two significant bits.
+% Results are summarized in Table~\ref{table:quality}.
+% In this table, STC(7) stands for embedding data in the LSB whereas
+% in STC(6), data are hidden in the last two significant bits.
-Let us give an interpretation of these experiments.
-First of all, the Adaptive strategy produces images with lower distortion
-than the images resulting from the 10\% fixed strategy.
-Numerical results are indeed always greater for the former strategy than
-for the latter one.
-These results are not surprising since the Adaptive strategy aims at
-embedding messages whose length is decided according to a higher threshold
-into the edge detection.
+% Let us give an interpretation of these experiments.
+% First of all, the Adaptive strategy produces images with lower distortion
+% than the images resulting from the 10\% fixed strategy.
+% Numerical results are indeed always greater for the former strategy than
+% for the latter one.
+% These results are not surprising since the Adaptive strategy aims at
+% embedding messages whose length is decided according to a higher threshold
+% into the edge detection.
-If we combine Adaptive and STC strategies
-the STABYLO scheme provides images whose quality is higher than
-the EAISLSBMR's one but lower than the quality of high complexity
-schemes. Notice that the quality of the less respectful scheme (EAILSBMR)
-is lower than 6\% than the one of the most one.
+% If we combine Adaptive and STC strategies
+% the STABYLO scheme provides images whose quality is higher than
+% the EAISLSBMR's one but lower than the quality of high complexity
+% schemes. Notice that the quality of the less respectful scheme (EAILSBMR)
+% is lower than 6\% than the one of the most one.
-% Let us now compare the STABYLO approach with other edge based steganography
-% approaches, namely~\cite{DBLP:journals/eswa/ChenCL10,Chang20101286}.
-% These two schemes focus on increasing the
-% payload while the PSNR is acceptable, but do not
-% give quality metrics for fixed embedding rates from a large base of images.
+% % Let us now compare the STABYLO approach with other edge based steganography
+% % approaches, namely~\cite{DBLP:journals/eswa/ChenCL10,Chang20101286}.
+% % These two schemes focus on increasing the
+% % payload while the PSNR is acceptable, but do not
+% % give quality metrics for fixed embedding rates from a large base of images.
Ensemble Classifier~\cite{DBLP:journals/tifs/KodovskyFH12} based steganalyser.
Its particularization to spatial domain is
considered as state of the art steganalysers.
-Firstly, a space
-of 686 co-occurrence and Markov features is extracted from the
-set of cover images and the set of training images. Next a small
+\JFC{Features that are embedded into this steganalysis process
+are CCPEV and SPAM features as described
+in~\cite{DBLP:dblp_conf/mediaforensics/KodovskyPF10}.
+These latter are extracted from the
+set of cover images and the set of training images.}
+Next a small
set of weak classifiers is randomly built,
each one working on a subspace of all the features.
The final classifier is constructed by a majority voting
However by combining Adaptive and STC strategies
our approach obtains similar results than the ones of these schemes.
-Compared to EAILSBMR, we obtain better results when the strategy is
+Compared to EAILSBMR, we obtain similar
+results when the strategy is
Adaptive.
-However due to its
-huge number of integration features, it is not lightweight.
+However due to its huge number of integration features, it is not lightweight.
All these numerical experiments confirm
the objective presented in the motivations:
providing an efficient steganography approach in a lightweight manner.
+
+\RC{In Figure~\ref{fig:error}, Ensemble Classifier has been used with all the previsou steganalizers with 3 different payloads. It can be observed that with important payload, STABYLO is not efficient, but as mentionned its complexity is far more simple compared to other tools.\\
+\begin{figure}
+\begin{center}
+\includegraphics[scale=0.5]{error}
+\end{center}
+\caption{Error obtained by Ensemble classifier with WOW/UNIWARD, HUGO, and STABYLO and different paylaods.}
+\label{fig:error}
+\end{figure}
+}
