--- scribo/ChangeLog | 4 + scribo/doc/research.tex | 230 +++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 234 insertions(+), 0 deletions(-) create mode 100644 scribo/doc/research.tex diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 0412e63..fd8acad 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,7 @@ +2011-05-18 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; + + * doc/research.tex: New file describing tests and conclusions. + 2011-05-17 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; Add a new tool. diff --git a/scribo/doc/research.tex b/scribo/doc/research.tex new file mode 100644 index 0000000..86ab68c --- /dev/null +++ b/scribo/doc/research.tex @@ -0,0 +1,230 @@ +%% Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +%% +%% This file is part of Olena. +%% +%% Olena is free software: you can redistribute it and/or modify it under +%% the terms of the GNU General Public License as published by the Free +%% Software Foundation, version 2 of the License. +%% +%% Olena is distributed in the hope that it will be useful, +%% but WITHOUT ANY WARRANTY; without even the implied warranty of +%% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +%% General Public License for more details. +%% +%% You should have received a copy of the GNU General Public License +%% along with Olena. If not, see <http://www.gnu.org/licenses/>. + +\documentclass[a4]{book} + +%\usepackage{hevea} + +\usepackage{html} +\usepackage{hyperref} +\usepackage{graphicx} +\usepackage{makeidx} +\usepackage{xcolor} +\usepackage{color} + +\title{SCRIBO\\ + \large{Research report} } +\author{LRDE} +\date{} +\makeindex + + +\begin{document} + +\maketitle + + + +%=========================================== +%=========================================== +%=========================================== +\chapter{Preprocessing} + + + +%******************************************* +%******************************************* +\section{Show-through removal} + + +%******************************************* +%******************************************* +\section{Color to grayscale conversion} + +2 formulas tested : +\begin{itemize} +\item $R + G + V$ +\item $0.299 * R + 0.587 * G + 0.114 * B$ +\end{itemize} + + +%******************************************* +%******************************************* +\section{Binarization} + + + +%........................................... +\subsection{Sauvola} +\par{Sauvola} + +\cite{Sauvola} + +Best published method for documents. + +Parameters set up according to \cite{Badekas}. + +\par{Sauvola Multi-scale} + +Implemented with integral images. \cite{Faisal.integral_images} + +\par{Sauvola 3-channels} + + + +%******************************************* +%******************************************* +\section{Background/Foreground identification} + + + +%******************************************* +%******************************************* +\section{Unskew} + + + +%******************************************* +%******************************************* +\section{Denoising} + + + +%******************************************* +%******************************************* +\section{Delimitors} + +%........................................... +\subsection{Lines} + +%........................................... +\subsection{Tab-stops and whitespaces} + +File concerned : scribo/primitive/extract/separators\_non\_visible.hh + +First attempt to retrieve tab-stops/whitespaces delimitors. In order +to limit false positive, the components are dilated horizontaly prior +the algorithm. + +False positive were still too numerous in the core paragraphes. + + +File concerned : scribo/primitive/extract/alignments.hh + +In order to avoid too much false positive, the text is grouped once +(almost by word). To limit connections between paragraphs, the rules +used to connect components is as follows : lookup for the closest left +neighbor until a maximum distance compute with the formula (w / 2.0f) ++ (dmax_factor_ * h), where w and h are respectively the width and the +height of the component. dmax_factor_ is a user defined parameter set +to 1. Functor primitive::link::internal::dmax_default is used and +implement that rule.. + +We tried to find tabstops and whitespaces without grouping first but +there were too much false positive inside paragraphs. Grouping may be +a problem some times since if two paragraphs are too close to +eachother, they may already connect... + + +%=========================================== +%=========================================== +%=========================================== +\chapter{Text extraction} + +%******************************************* +%******************************************* +\section{lines} + +%........................................... +\subsection{Component labeling} + +%........................................... +\subsection{Component grouping} + +%........................................... +\subsection{Line reconstruction} + + + +%******************************************* +%******************************************* +\section{paragraphs/text blocks} + + +%=========================================== +%=========================================== +%=========================================== +\chapter{Non-text object extraction} + +%******************************************* +%******************************************* +\section{Background learning} + + +%=========================================== +%=========================================== +%=========================================== +\chapter{Text recognition (OCR)} + +%******************************************* +%******************************************* +\section{Tesseract Integration} + + +%******************************************* +%******************************************* +\section{Text cleanup} + + +%=========================================== +%=========================================== +%=========================================== +\chapter{Data structures} + +%******************************************* +%******************************************* +\section{Component\_set} +\subsection{Component\_info} + +%******************************************* +%******************************************* +\section{object\_links} + +%******************************************* +%******************************************* +\section{object\_groups} + + + +%******************************************* +%******************************************* +\section{line\_set} + +%........................................... +\subsection{line\_info} + + + + +%******************************************* +%******************************************* +\section{paragraph\_set} + +%........................................... +\subsection{paragraph\_info} + +\end{document} + -- 1.5.6.5