From 01ec38902455e4ee4dba09ead751639638481fd2 Mon Sep 17 00:00:00 2001 From: couchot Date: Fri, 14 Nov 2014 10:20:06 +0100 Subject: [PATCH] reprise intro --- experiments.tex | 2 +- intro.tex | 27 ++++++++++++++++++++++----- main.tex | 11 +++++++++-- ourapproach.tex | 28 ++++++++++++++++++++-------- 4 files changed, 52 insertions(+), 16 deletions(-) diff --git a/experiments.tex b/experiments.tex index 5ba43df..498e472 100644 --- a/experiments.tex +++ b/experiments.tex @@ -1,5 +1,5 @@ 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 $512\times 512$ diff --git a/intro.tex b/intro.tex index 4c99548..cd2aeea 100644 --- a/intro.tex +++ b/intro.tex @@ -57,7 +57,7 @@ Thus a distortion measure for each pixel is individually determined as the sum of the differences between the features of the SPAM computed from the cover and from the stego images. The features embedded in WOW and UNIWARD are based on Wavelet-based -directional filter. Thus, similarelly, the distortion function is +directional filter. Thus, similarly, the distortion function is the sum of the differences between these wavelet coefficients computed from the cover and from the stego images. @@ -108,10 +108,27 @@ One can notice that all the previously referenced sche\-mes~\cite{Luo:2010:EAI:1824719.1824720,DBLP:journals/eswa/ChenCL10,DBLP:conf/ih/PevnyFB10} produce stego contents by only considering the payload, not the type of image signal: the higher the payload is, -the better the approach is said to be. -Contrarily, we argue that some images should not be taken as a cover because of the nature of their signals. -Consider for instance a uniformly black image: a very tiny modification of its pixels can be easily detectable. -The approach we propose is thus to provide a self adaptive algorithm with a high payload, which depends on the cover signal. +the better the approach is said to be. +For instance, studied payloads range from 0.04 to 0.4 modified bits per pixel. +Contrarily, we argue that some images should not be taken +as a cover because of the nature of their signals. +Consider for instance a uniformly black image: a very tiny modification of its +pixels can be easily detectable. +Practically speaking, if Alice would send +a hidden message to Bob, she would never consider +such kind of image and a high embedding rate. +The approach we propose here is thus to provide a small complexity +self adaptive algorithm +with an acceptable payload, which +depends on the cover signal. +The payload is further said to + be acceptable if it allows to embed a sufficiently +long message in the cover signal. +Practically speaking, our approach is efficient enough for +payloads close to 0.06 bit per pixel which allows to embed +messages of length larger than 16000 bits in an +image of size $512\times 512$ pixels. + % Message extraction is achieved by computing the same % edge detection pixels set for the cover and the stego image. % The edge detection algorithm is thus not applied on all the bits of the image, diff --git a/main.tex b/main.tex index e6405e6..e82040e 100755 --- a/main.tex +++ b/main.tex @@ -147,8 +147,15 @@ to face steganalysers. For future work, the authors' intention is to investigate systematically -all the existing edge detection methods, to see if the STABYLO evaluation scores can -be improved by replacing Canny with another edge filter. +all the existing edge detection methods, +to see if the STABYLO evaluation scores can +be improved by replacing Canny with another edge filter. +Moreover, we plan to improve the distortion function by integrating +into a numerical cost the gradient value of this kind of +algorithm. We could thus transmit this value to STC contrary to the current +version where the distortion that is transmited is either 1 in +the adaptive strategy or 1,10, 100 in the fixed strategy. + Other steganalysers than the ones used in this document will be examined for the sake of completeness. Finally, the systematic replacement of all the LSBs of edges by binary digits provided diff --git a/ourapproach.tex b/ourapproach.tex index 12b3a32..3537640 100644 --- a/ourapproach.tex +++ b/ourapproach.tex @@ -196,14 +196,26 @@ It is further referred to as \emph{STC} and is detailed in the next section. % but the whole approach can be updated to consider % the fuzzy logic edge detector. -% Next, following~\cite{Luo:2010:EAI:1824719.1824720}, our scheme automatically -% modifies Canny parameters to get a sufficiently large set of edge bits: this -% one is practically enlarged untill its size is at least twice as many larger -% than the size of embedded message. - - - -%%RAPH: paragraphe en double :-) +For a given set of parameters, +the canny algorithm returns a numerical value and +states whether a given pixel is an edge or not. +In this article, in the Adaptive strategy +we consider that all the edge pixels that +have been selected by this algorithm have the same +distortion cost \textit{i.e.} $\rho_X$ is always 1 for these bits. +In the Fixed strategy, since pixels that are detected to be edge +with small values of $T$ (e.g. when $T=3$) +are more accurate than these with higher values of $T$, +we give to STC the following distortion map of the corresponding bits +$$ +\rho_X= \left\{ +\begin{array}{l} +1 \textrm{ if an edge for $T=3$} \\ +10 \textrm{ if an edge for $T=5$} \\ +100 \textrm{ if an edge for $T=7$} +\end{array} +\right. +$$. -- 2.39.5