Code: Select all
\documentclass[12pt,journal,compsoc]{IEEEtran}
\usepackage{alltt}
\usepackage{amssymb}
\usepackage{amsfonts}
\usepackage{mathtools}
\usepackage{array}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage{float}
\usepackage{lipsum}
\begin{document}
\subsection{Related Work}
Fractal encoding can be done in spatial domain[10][11] and frequency domain[12]. Paul Darvin[13] suggested a fractal method that uses the information of range and domain region as key. Joan Paute and Fred Jordan[14] used key to generate coordinates of range blocks. Patrick Bas and Jean-Marc Chassery[15][16] used fractal scheme for watermarking. They find the transformation such that for each block D and R, a new range block $\hat R$ is calculated as\\
$\hat R = \frac{\delta*S*D}{max(D)}+\bar R$\\
where S determines the magnitude of the watermark, $\bar R$ is the mean of R and
\[ \delta = \left\{
\begin{array}{l l}
& \quad \text{+1 if the embedded bit=1}\\
& \quad \text{ -1 if the embedded bit=0}
\end{array} \right.
\]
\subsection{Implementation}
Implementation of the algorithm is based on the following assumptions:
\begin{table}[ht]
\centering % used for centering table
\begin{tabular}{l} % centered columns (1 columns)
\hline % inserts single horizontal line
\\
\noindent
$\bullet$ Append data with the label END OF DATA.\\
$\bullet$ XOR data with key.\\
$\bullet$ Subsample domain blocks D so that they have same number of pixels \\
as range blocks.\\
$\bullet$ Classify all domain blocks.\\
$\bullet$ While there are bits to be stored \\
\{\\
$\bullet$ Take a range block $R_i$ from R. \\
$\bullet$ Classify $R_i$. \\
$\bullet$ If current bit bi is 0 \\
$\bullet$ Consider domain blocks from $D^0$ that are of same class as of $R_i$. \\
$\bullet$ else \\
$\bullet$ Consider domain blocks from $D^1$ that are of same class as of $R_i$. \\
$\bullet$ Calculate scale $s_i$, offset $o_i$ and rms distance $d_{rms}$ between $R_i$ and all \\
•$D_j$s in the selected quadrant. \\
$\bullet$ Select domain block $D_min$ with least $d_{rms}$ and corresponding $s_i$ and \\
• $o_i$.\\
$\bullet$ Multiply each pixel of $D_min$ by $s_i$ and add $o_i$ and overwrite the \\
•corresponding range pixel by the result. \\
\}\\
\\
\hline %inserts single line
\end{tabular}
\label{table:nonlin} % is used to refer this table in the text
\caption{Pseudo-code for hiding steganographic data.} % title of Table
\end{table}
\begin{table}[ht]
\centering % used for centering table
\begin{tabular}{l} % centered columns (1 columns)
\hline % inserts single horizontal line
\\
$\bullet$ While not END OF DATA \\
\{\\
$\bullet$ Take a range block $R_i$ from R.\\
$\bullet$ Calculate scale $s_i$, offset $o_i$ and rms distance drms between $R_i$ and all\\
\'Djs in D.\\
$\bullet$ Select domain block $D_{min}$ with least $d_{rms}$ and corresponding $s_i$ \\
and $o_i$.\\
$\bullet$ If Dj is in $D^0$. \\
$\bullet$ Data bit bi is 0. \\
$\bullet$ else \\
$\bullet$ Data bit bi is 1 \\
\}\\
$\bullet$ XOR this bit stream with key to get original data.\\
\\
\hline %inserts single line
\end{tabular}
\label{table:nonlin} % is used to refer this table in the text
\caption{Pseudo-code for retrieving steganographic data.} % title of Table
\end{table}
\noindent
$\bullet$ It is assumed that the input image is of equal length and breadth.\\
$\bullet$ Image is grey scale and in the sun-raster format.\\
$\bullet$ Size of image header is 800 bytes.\\
$\bullet$ Each byte in the image data represents a pixel whose level of grey is from 0 to 255.\\
The hiding algorithm takes image I
\end{document}
\end{document}
After 2 tables the bullet-ed lines must come.But it is not acting so.
