+First of all, the whole code of STABYLO can be downloaded
+\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 $512\times 512$
+grayscale digital image in a RAW format.
+We restrict experiments to
+this set of cover images since this paper is more focused on
+the methodology than on benchmarks.
-\subsection{Image Quality}
-The visual quality of the STABYLO scheme is evaluated in this section.
-Three metrics are computed in these experiments :
-the Peak Signal to Noise Ratio (PSNR),
-the PSNR-HVS-M~\cite{PSECAL07,psnrhvsm11} and the BIQI~\cite{MB10,biqi11}.
-The first one is widely used but does not take into
-account Human Visual System (HVS).
-The two last ones have been designed to tackle this problem.
+We use the matrices $\hat{H}$
+generated by the integers given
+in Table~\ref{table:matrices:H}
+as introduced in~\cite{FillerJF11}, since these ones have experimentally
+be proven to have the strongest modification efficiency.
+For instance if the rate between the size of the message and the size of the
+cover vector
+is 1/4, each number in $\{81, 95, 107, 121\}$ is translated into a binary number
+and each one constitutes thus a column of $\hat{H}$.
+\begin{table}
+$$
+\begin{array}{|l|l|}
+\hline
+\textrm{Rate} & \textrm{Matrix generators} \\
+\hline
+{1}/{2} & \{71,109\}\\
+\hline
+{1}/{3} & \{95, 101, 121\}\\
+\hline
+{1}/{4} & \{81, 95, 107, 121\}\\
+\hline
+{1}/{5} & \{75, 95, 97, 105, 117\}\\
+\hline
+{1}/{6} & \{73, 83, 95, 103, 109, 123\}\\
+\hline
+{1}/{7} & \{69, 77, 93, 107, 111, 115, 121\}\\
+\hline
+{1}/{8} & \{69, 79, 81, 89, 93, 99, 107, 119\}\\
+\hline
+{1}/{9} & \{69, 79, 81, 89, 93, 99, 107, 119, 125\}\\
+\hline
+\end{array}
+$$
+\caption{Matrix Generator for $\hat{H}$ in STC.}\label{table:matrices:H}
+\end{table}
-biqi = 28.3
-psnr-hvs-m= 78,6
-psnr-hvs= 67.3
+Our approach is always compared to HUGO, to EAISLSBMR, to WOW and to UNIWARD
+for the two strategies Fixed and Adaptive.
+For the former one, the payload has been set to 10\%.
+For the latter one, the Canny parameter $T$ has been set to 3.
+When $b$ is 7, the average size of the message that can be embedded
+is 16,445 bits,
+that corresponds to an average payload of 6.35\%.
+For each cover image the STABYLO's embedding rate with these two parameters
+is memorized.
+Next each steganographic scheme is executed to produce the stego content of
+this cover with respect to this embedding rate.
-\subsection{Steganalysis}
+
+% \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.
-The quality of our approach has been evaluated through the two
-AUMP~\cite{Fillatre:2012:ASL:2333143.2333587}
-and Ensemble Classifier~\cite{DBLP:journals/tifs/KodovskyFH12} based steganalysers.
-Both aims at detecting hidden bits in grayscale natural images and are
-considered as the state of the art of steganalysers in spatial domain~\cite{FK12}.
-The former approach is based on a simplified parametric model of natural images.
-Parameters are firstly estimated and a adaptive Asymptotically Uniformly Most Powerful
-(AUMP) test is designed (theoretically and practically) to check whether
-a natural image has stego content or not.
-In the latter, the authors show that the
-machine learning step, (which is often
-implemented as support vector machine)
-can be a favourably executed thanks to an Ensemble Classifiers.
+% 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.
-\JFC{Raphael, il faut donner des résultats ici}
\ No newline at end of file
+
+% \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.
+
+
+% 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.
+
+
+% % 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.
+
+
+
+
+\subsection{Steganalysis}\label{sub:steg}
+
+
+
+The steganalysis quality of our approach has been evaluated through the % two
+% AUMP~\cite{Fillatre:2012:ASL:2333143.2333587}
+% and
+Ensemble Classifier~\cite{DBLP:journals/tifs/KodovskyFH12} based steganalyser.
+Its particularization to spatial domain is
+considered as state of the art steganalysers.
+\JFC{Features that are embedded into this steganalysis process
+are CCPEV and SPAM features as described
+in~\cite{DBLP:dblp_conf/mediaforensics/KodovskyPF10}.
+They 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
+between the decisions of these individual classifiers.
+
+
+%The former approach is based on a simplified parametric model of natural images.
+% Parameters are firstly estimated and an adaptive Asymptotically Uniformly Most Powerful
+% (AUMP) test is designed (theoretically and practically), to check whether
+% an image has stego content or not.
+% This approach is dedicated to verify whether LSB has been modified or not.
+% , the authors show that the
+% machine learning step, which is often
+% implemented as a support vector machine,
+% can be favorably executed thanks to an ensemble classifier.
+
+
+\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 & Adapt. & Fixed & Adapt. & Fixed & Adapt. & Fixed & Adapt. \\
+\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
+%AUMP & 0.22 & 0.33 & 0.39 & 0.45 & 0.50 & 0.50 & 0.49 & 0.50 \\
+%\hline
+Ensemble Classifier & 0.35 & 0.44 & 0.47 & 0.47 & 0.48 & 0.49 & 0.43 & 0.47 & 0.48 & 0.49 & 0.46 & 0.49 \\
+
+\hline
+\end{tabular}
+\end{small}
+\end{center}
+\caption{Steganalysing STABYLO\label{table:steganalyse}.}
+\end{table*}
+
+
+Results of average testing errors
+are summarized in Table~\ref{table:steganalyse}.
+First of all, STC outperforms the sample strategy %for % the two steganalysers
+ as
+already noticed in the quality analysis presented in the previous section.
+Next, our approach is more easily detectable than HUGO,
+WOW and UNIWARD which are the most secure steganographic tool,
+as far as we know.
+However by combining Adaptive and STC strategies
+our approach obtains similar results than the ones of these schemes.
+
+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.
+
+All these numerical experiments confirm
+the objective presented in the motivations:
+providing an efficient steganography approach in a lightweight manner
+for small payload.
+
+\RC{In Figure~\ref{fig:error},
+Ensemble Classifier has been used with all the previous
+steganographic schemes with 4 different payloads.
+It can be observed that face to high values of payload,
+STABYLO is definitely not secure enough.
+However thanks to an efficient very low-complexity (Fig.\ref{fig:compared}),
+we argue that the user should embed tiny messages in many images
+than a larger message in only one image.
+\begin{figure}
+\begin{center}
+\includegraphics[scale=0.5]{error}
+\end{center}
+\caption{Testing error obtained by Ensemble classifier with
+WOW/UNIWARD, HUGO, and STABYLO w.r.t. payload.}
+\label{fig:error}
+\end{figure}
+}
