X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/canny.git/blobdiff_plain/7aeb1ab6cc41dd51a6dbe4c844c8983809205186..d8baf5b58b6fba7154744aa204a10b8530c0d2a9:/experiments.tex diff --git a/experiments.tex b/experiments.tex index eff125e..15fb5c1 100644 --- a/experiments.tex +++ b/experiments.tex @@ -42,7 +42,7 @@ $$ \hline \end{array} $$ -\caption{Matrix Generator for $\hat{H}$ in STC}\label{table:matrices:H} +\caption{Matrix Generator for $\hat{H}$ in STC.}\label{table:matrices:H} \end{table} @@ -62,80 +62,80 @@ this cover with respect to this embedding rate. -\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. @@ -150,9 +150,12 @@ The steganalysis quality of our approach has been evaluated through the % two 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}. +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 @@ -190,7 +193,7 @@ Ensemble Classifier & 0.35 & 0.44 & 0.47 & 0.47 & 0.48 & 0.49 & 0.43 & 0 \end{tabular} \end{small} \end{center} -\caption{Steganalysing STABYLO\label{table:steganalyse}} +\caption{Steganalysing STABYLO\label{table:steganalyse}.} \end{table*} @@ -212,4 +215,23 @@ 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. +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} +}