[Olena-patches] last-svn-commit-746-g9549f83 Add paragraph processing in content_in_doc toolchain.

* scribo/toolchain/content_in_doc.hh: New. * scribo/toolchain/internal/content_in_doc_functor.hh: Use paragraph related routines. * src/content_in_doc.cc: Update use of content_in_doc_functor. --- scribo/ChangeLog | 11 ++ .../{text_in_doc.hh => content_in_doc.hh} | 52 +++++----- .../toolchain/internal/content_in_doc_functor.hh | 106 +++++++++++++++++++- scribo/src/content_in_doc.cc | 48 +++------- 4 files changed, 155 insertions(+), 62 deletions(-) copy scribo/scribo/toolchain/{text_in_doc.hh => content_in_doc.hh} (57%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index c947550..809ce7d 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,14 @@ +2011-01-26 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; + + Add paragraph processing in content_in_doc toolchain. + + * scribo/toolchain/content_in_doc.hh: New. + + * scribo/toolchain/internal/content_in_doc_functor.hh: Use + paragraph related routines. + + * src/content_in_doc.cc: Update use of content_in_doc_functor. + 2011-01-25 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; Small fixes in Scribo. diff --git a/scribo/scribo/toolchain/text_in_doc.hh b/scribo/scribo/toolchain/content_in_doc.hh similarity index 57% copy from scribo/scribo/toolchain/text_in_doc.hh copy to scribo/scribo/toolchain/content_in_doc.hh index e6ba69e..f2938d9 100644 --- a/scribo/scribo/toolchain/text_in_doc.hh +++ b/scribo/scribo/toolchain/content_in_doc.hh @@ -1,5 +1,4 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -24,14 +23,14 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -#ifndef SCRIBO_TOOLCHAIN_TEXT_IN_DOC_HH -# define SCRIBO_TOOLCHAIN_TEXT_IN_DOC_HH +#ifndef SCRIBO_TOOLCHAIN_CONTENT_IN_DOC_HH +# define SCRIBO_TOOLCHAIN_CONTENT_IN_DOC_HH /// \file /// -/// Extract text from a document. +/// Analyse a document. -# include <scribo/toolchain/internal/text_in_doc_functor.hh> +# include <scribo/toolchain/internal/content_in_doc_functor.hh> namespace scribo { @@ -42,36 +41,41 @@ namespace scribo using namespace mln; - template <typename I> - line_set<mln_ch_value(I, def::lbl_type)> - text_in_doc(const Image<I>& input, bool denoise, - const std::string& language = std::string("eng"), - bool find_line_seps = true, - bool find_whitespace_seps = true, - bool debug = false); + template <typename I, typename J> + document<mln_ch_value(I, def::lbl_type)> + content_in_doc(const Image<I>& input, const Image<J>& input_preproc, + bool denoise, + const std::string& language = std::string("eng"), + bool find_line_seps = true, + bool find_whitespace_seps = true, + bool debug = false); # ifndef MLN_INCLUDE_ONLY - template <typename I> - line_set<mln_ch_value(I, def::lbl_type)> - text_in_doc(const Image<I>& input, bool denoise, - const std::string& language = std::string("eng"), - bool find_line_seps = true, - bool find_whitespace_seps = true, - bool debug = false) + template <typename I, typename J> + document<mln_ch_value(I, def::lbl_type)> + content_in_doc(const Image<I>& input, const Image<J>& input_preproc, + bool denoise, + const std::string& language = std::string("eng"), + bool find_line_seps = true, + bool find_whitespace_seps = true, + bool debug = false) { - internal::text_in_doc_functor<I> f; + mln_precondition(input.is_valid()); + mln_precondition(input_preproc.is_valid()); + + internal::content_in_doc_functor<J> f("noname"); f.enable_denoising = denoise; f.enable_line_seps = find_line_seps; f.enable_whitespace_seps = find_whitespace_seps; f.enable_debug = debug; f.ocr_language = language; - line_set<mln_ch_value(I, def::lbl_type)> lines = f(input); + document<mln_ch_value(I, def::lbl_type)> doc = f(input, input_preproc); - return lines; + return doc; } @@ -83,5 +87,5 @@ namespace scribo } // end of namespace scribo -#endif // SCRIBO_TOOLCHAIN_TEXT_IN_DOC_HH +#endif // SCRIBO_TOOLCHAIN_CONTENT_IN_DOC_HH diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index 12e5137..7c665e5 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -30,14 +30,20 @@ # include <scribo/core/def/lbl_type.hh> # include <scribo/core/document.hh> +# include <scribo/core/line_set.hh> +# include <scribo/core/paragraph_set.hh> # include <scribo/primitive/extract/elements.hh> # include <scribo/primitive/extract/components.hh> # include <scribo/primitive/extract/vertical_separators.hh> # include <scribo/primitive/extract/separators_nonvisible.hh> +# include <scribo/primitive/extract/elements.hh> + +# include <scribo/primitive/identify.hh> # include <scribo/primitive/remove/separators.hh> +# include <scribo/filter/line_links_x_height.hh> # include <scribo/filter/object_links_bbox_h_ratio.hh> # include <scribo/filter/objects_small.hh> @@ -52,6 +58,7 @@ # include <scribo/text/recognition.hh> # include <scribo/text/merging.hh> +# include <scribo/text/link_lines.hh> # include <scribo/make/debug_filename.hh> @@ -336,6 +343,7 @@ namespace scribo lines = scribo::text::merging(lines); + //===== DEBUG ===== if (enable_debug) { @@ -353,24 +361,116 @@ namespace scribo } + //===== END OF DEBUG ===== on_progress(); - on_new_progress_label("Recognizing text"); + // Text recognition + on_new_progress_label("Recognizing text"); scribo::text::recognition(lines, ocr_language.c_str()); - doc.set_text(lines); on_progress(); + + // Link text lines + on_new_progress_label("Linking text lines"); + line_links<L> llinks = scribo::text::link_lines(lines); + + + //===== DEBUG ===== + if (enable_debug) + { + image2d<value::rgb8> debug = data::convert(value::rgb8(), original_image); + for_all_lines(l, lines) + { + if (! lines(l).is_valid() || lines(l).is_hidden() || lines(l).type() != line::Text) + continue; + + mln::draw::box(debug, lines(l).bbox(), literal::blue); + mln::draw::line(debug, lines(l).bbox().pcenter(), lines(llinks(l)).bbox().pcenter(), literal::green); + } + + mln::io::ppm::save(debug, scribo::make::debug_filename("links_raw.ppm")); + } + //===== END OF DEBUG ===== + + on_progress(); + + + // Filter line links. + on_new_progress_label("Filter line links"); + llinks = scribo::filter::line_links_x_height(llinks); + + //===== DEBUG ===== + if (enable_debug) + { + image2d<value::rgb8> debug = data::convert(value::rgb8(), original_image); + for_all_links(i, llinks) + if (llinks(i) && llinks(i) != i) + mln::draw::line(debug, lines(i).bbox().pcenter(), + lines(llinks(i)).bbox().pcenter(), literal::red); + + mln::io::ppm::save(debug, scribo::make::debug_filename("links.ppm")); + + + for (unsigned i = 1; i < llinks.nelements(); ++i) + llinks(i) = scribo::make::internal::find_root(llinks, i); + + debug = data::convert(value::rgb8(), original_image); + mln::util::array<accu::shape::bbox<point2d> > nbbox(llinks.nelements()); + for_all_lines(i, lines) + { + if (! lines(i).is_valid() || lines(i).is_hidden() || lines(i).type() != line::Text) + continue; + + mln::draw::box(debug, lines(i).bbox(), literal::red); + nbbox(llinks(i)).take(lines(i).bbox()); + } + + for (unsigned i = 1; i < nbbox.nelements(); ++i) + if (nbbox(i).is_valid()) + { + box2d b = nbbox(i).to_result(); + mln::draw::box(debug, b, literal::green); + b.enlarge(1); + mln::draw::box(debug, b, literal::green); + b.enlarge(1); + mln::draw::box(debug, b, literal::green); + } + + mln::io::ppm::save(debug, scribo::make::debug_filename("par.ppm")); + } + //===== END OF DEBUG ===== + + on_progress(); + + + // Construct paragraphs + on_new_progress_label("Constructing paragraphs"); + scribo::paragraph_set<L> parset = scribo::make::paragraph(llinks); + doc.set_paragraphs(parset); + + on_progress(); + + // Extract other Elements on_new_progress_label("Extracting Elements"); component_set<L> elements = scribo::primitive::extract::elements(doc, original_image); + + on_progress(); + + + // Identify other Elements + on_new_progress_label("Identifying Elements"); + elements = scribo::primitive::identify(elements); doc.set_elements(elements); on_progress(); + + // Saving results if (save_doc_as_xml) { @@ -391,7 +491,7 @@ namespace scribo int content_in_doc_functor<I>::nsteps() const { - return 7 + enable_denoising + enable_line_seps + return 11 + enable_denoising + enable_line_seps + enable_whitespace_seps + save_doc_as_xml; } diff --git a/scribo/src/content_in_doc.cc b/scribo/src/content_in_doc.cc index f453f08..8cd262b 100644 --- a/scribo/src/content_in_doc.cc +++ b/scribo/src/content_in_doc.cc @@ -30,38 +30,23 @@ #include <iostream> #include <mln/core/image/image2d.hh> -#include <mln/core/alias/neighb2d.hh> #include <mln/io/pbm/save.hh> #include <mln/io/magick/load.hh> -#include <mln/value/label_8.hh> - -#include <mln/core/var.hh> - -#include <mln/accu/count_value.hh> - -#include <mln/draw/box_plain.hh> - - -#include <scribo/toolchain/text_in_doc.hh> +#include <scribo/toolchain/content_in_doc.hh> #include <scribo/toolchain/text_in_doc_preprocess.hh> #include <scribo/core/document.hh> -#include <scribo/core/line_set.hh> #include <scribo/debug/usage.hh> #include <scribo/make/debug_filename.hh> -#include <scribo/primitive/extract/elements.hh> - #include <scribo/preprocessing/crop_without_localization.hh> #include <scribo/preprocessing/crop.hh> #include <scribo/io/xml/save.hh> -#include <scribo/io/text_boxes/save.hh> - const char *args_desc[][2] = @@ -108,11 +93,11 @@ int main(int argc, char* argv[]) Magick::InitializeMagick(*argv); typedef image2d<scribo::def::lbl_type> L; - scribo::document<L> doc(argv[1]); - doc.open(); + image2d<value::rgb8> input; + mln::io::magick::load(input, argv[1]); // Preprocess document - image2d<bool> input; + image2d<bool> input_preproc; { double K = 0.34; if (argc == 8 || argc == 12 || argc >= 12) @@ -125,7 +110,7 @@ int main(int argc, char* argv[]) } image2d<bool> tmp_fg; - input = toolchain::text_in_doc_preprocess(doc.image(), false, K); + input_preproc = toolchain::text_in_doc_preprocess(input, false, K); } // Optional Cropping @@ -142,12 +127,12 @@ int main(int argc, char* argv[]) << " to (" << maxr << "," << maxc << ")" << std::endl; box2d roi = mln::make::box2d(minr, minc, maxr, maxc); - input = preprocessing::crop_without_localization(input, roi); + input_preproc = preprocessing::crop_without_localization(input_preproc, roi); crop_shift = point2d(minr, minc); if (debug) - mln::io::pbm::save(input, - scribo::make::debug_filename("input_cropped.pbm")); + mln::io::pbm::save(input_preproc, + scribo::make::debug_filename("input_preproc_cropped.pbm")); } bool denoise = (argc > 3 && atoi(argv[3]) != 0); @@ -174,18 +159,11 @@ int main(int argc, char* argv[]) // Run document toolchain. // Text - std::cout << "Extracting text" << std::endl; - line_set<L> - lines = scribo::toolchain::text_in_doc(input, denoise, language, - find_line_seps, find_whitespace_seps, - debug); - doc.set_text(lines); - - // Elements - std::cout << "Extracting Elements" << std::endl; - component_set<L> elements = scribo::primitive::extract::elements(doc, input); - doc.set_elements(elements); - + std::cout << "Analysing document..." << std::endl; + document<L> + doc = scribo::toolchain::content_in_doc(input, input_preproc, denoise, language, + find_line_seps, find_whitespace_seps, + debug); // Saving results scribo::io::xml::save(doc, argv[2], true); -- 1.5.6.5