- Olena-patches

branch fix-leak-remove_holder updated: last-svn-commit-911-g49106f5
by lazzara＠lrde.epita.fr 02 Jun '11

02 Jun '11

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "Olena, a generic and efficient image processing platform". The branch fix-leak-remove_holder has been updated via 49106f51b4957bc4af897c365f0458124591f552 (commit) from 4543e118e2559cbd15afd3bf26531d1db6dde9d3 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- 49106f5 Add a toolchain for DAE contest. ----------------------------------------------------------------------- Summary of changes: .../toolchain/internal/content_in_doc_functor.hh | 5 +++++ scribo/src/Makefile.am | 3 ++- scribo/src/contest/DAE-2011/Makefile.am | 11 ++++++++++- scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc | 4 ++-- 4 files changed, 19 insertions(+), 4 deletions(-) hooks/post-receive -- Olena, a generic and efficient image processing platform

1 0

last-svn-commit-911-g49106f5 Add a toolchain for DAE contest.
by Guillaume Lazzara 02 Jun '11

02 Jun '11

--- .../toolchain/internal/content_in_doc_functor.hh | 5 +++++ scribo/src/Makefile.am | 3 ++- scribo/src/contest/DAE-2011/Makefile.am | 11 ++++++++++- scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc | 4 ++-- 4 files changed, 19 insertions(+), 4 deletions(-) diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index 4308056..4f2c074 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -57,7 +57,10 @@ # include <scribo/preprocessing/denoise_fg.hh> +# ifndef SCRIBO_NOCR # include <scribo/text/recognition.hh> +# endif // ! SCRIBO_NOCR + # include <scribo/text/merging.hh> # include <scribo/text/link_lines.hh> # include <scribo/text/paragraphs.hh> @@ -437,6 +440,7 @@ namespace scribo on_progress(); +# ifndef SCRIBO_NOCR // Text recognition if (enable_ocr) { @@ -446,6 +450,7 @@ namespace scribo on_progress(); } +# endif // ! SCRIBO_NOCR // // Link text lines // on_new_progress_label("Linking text lines"); diff --git a/scribo/src/Makefile.am b/scribo/src/Makefile.am index 6ab5d7d..26f2ade 100644 --- a/scribo/src/Makefile.am +++ b/scribo/src/Makefile.am @@ -28,7 +28,8 @@ SUBDIRS = \ preprocessing \ table \ text \ - toolchain + toolchain \ + util diff --git a/scribo/src/contest/DAE-2011/Makefile.am b/scribo/src/contest/DAE-2011/Makefile.am index 09bccd7..d7d6b8d 100644 --- a/scribo/src/contest/DAE-2011/Makefile.am +++ b/scribo/src/contest/DAE-2011/Makefile.am @@ -24,7 +24,7 @@ noinst_PROGRAMS = if HAVE_MAGICKXX -noinst_PROGRAMS += content_in_hdoc_dae +noinst_PROGRAMS += content_in_hdoc_dae content_in_doc_dae content_in_hdoc_dae_SOURCES = content_in_hdoc_dae.cc content_in_hdoc_dae_CPPFLAGS = $(AM_CPPFLAGS) \ @@ -34,4 +34,13 @@ content_in_hdoc_dae_LDFLAGS = $(AM_LDFLAGS) \ -lpthread \ $(MAGICKXX_LDFLAGS) +content_in_doc_dae_SOURCES = content_in_doc_dae.cc +content_in_doc_dae_CPPFLAGS = $(AM_CPPFLAGS) \ + $(TESSERACT_CPPFLAGS) \ + -I$(top_builddir) +content_in_doc_dae_LDFLAGS = $(AM_LDFLAGS) \ + -lpthread \ + $(MAGICKXX_LDFLAGS) + + endif HAVE_MAGICKXX \ No newline at end of file diff --git a/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc index f13b8f6..5986142 100644 --- a/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc +++ b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc @@ -1,4 +1,4 @@ -// Copyright (C) 2010, 2011 EPITA Research and Development Laboratory +// Copyright (C) 2011 EPITA Research and Development Laboratory // (LRDE) // // This file is part of Olena. @@ -24,7 +24,7 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -# define SCRIBO_NOCR +#define SCRIBO_NOCR #include <libgen.h> #include <fstream> -- 1.5.6.5

1 0

branch fix-leak-remove_holder updated: last-svn-commit-910-g4543e11
by lazzara＠lrde.epita.fr 01 Jun '11

01 Jun '11

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "Olena, a generic and efficient image processing platform". The branch fix-leak-remove_holder has been updated via 4543e118e2559cbd15afd3bf26531d1db6dde9d3 (commit) from 1dec414d7cbe485e888953444197cc645d076540 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- 4543e11 Specify output dir. ----------------------------------------------------------------------- Summary of changes: scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-) hooks/post-receive -- Olena, a generic and efficient image processing platform

1 0

last-svn-commit-910-g4543e11 Specify output dir.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

--- scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-) diff --git a/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc index 5309d49..f13b8f6 100644 --- a/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc +++ b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc @@ -60,6 +60,7 @@ const char *args_desc[][2] = { { "input.tif", "An image." }, + { "output_dir", "Output directory." }, {0, 0} }; @@ -70,10 +71,10 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 2) + if (argc != 3) return scribo::debug::usage(argv, "Document Image Analysis in Historical Documents", - "input.tif", + "input.tif output_dir", args_desc); trace::entering("main"); @@ -144,7 +145,7 @@ int main(int argc, char* argv[]) } std::stringstream ss; - ss << basename(argv[1]) << p << ".pbm"; + ss << argv[2] << "/" << basename(argv[1]) << "." << p << ".pbm"; mln::io::pbm::save(output, ss.str()); } -- 1.5.6.5

1 0

branch fix-leak-remove_holder created: last-svn-commit-909-g1dec414
by lazzara＠lrde.epita.fr 01 Jun '11

01 Jun '11

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "Olena, a generic and efficient image processing platform". The branch fix-leak-remove_holder has been created at 1dec414d7cbe485e888953444197cc645d076540 (commit) - Log ----------------------------------------------------------------- 1dec414 Add binaries for ICDAR contests. b0db028 Small fixes. a714708 Fix use of uninitialized values. e6c5288 Fix a serious memory leak. 7142b16 BACKUP bce6cef BACKUP ----------------------------------------------------------------------- hooks/post-receive -- Olena, a generic and efficient image processing platform

1 0

last-svn-commit-909-g1dec414 Add binaries for ICDAR contests.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

--- configure.ac | 3 + .../toolchain/internal/content_in_hdoc_functor.hh | 6 +- scribo/scribo/toolchain/text_in_doc_preprocess.hh | 1 + scribo/src/content_in_hdoc.cc | 20 ++-- .../estim => src/contest/DAE-2011}/Makefile.am | 22 +++-- .../DAE-2011/content_in_hdoc_dae.cc} | 111 ++++++++++++-------- scribo/src/contest/Makefile.am | 7 +- .../estim => src/contest/hdlac-2011}/Makefile.am | 22 +++-- .../hdlac-2011/content_in_hdoc_hdlac.cc} | 91 +++++++++-------- 9 files changed, 172 insertions(+), 111 deletions(-) copy scribo/{tests/estim => src/contest/DAE-2011}/Makefile.am (64%) copy scribo/src/{non_text_components.cc => contest/DAE-2011/content_in_hdoc_dae.cc} (51%) copy scribo/{tests/estim => src/contest/hdlac-2011}/Makefile.am (63%) copy scribo/src/{non_text_components.cc => contest/hdlac-2011/content_in_hdoc_hdlac.cc} (56%) diff --git a/configure.ac b/configure.ac index fdbddd9..66643a4 100644 --- a/configure.ac +++ b/configure.ac @@ -323,6 +323,8 @@ AC_CONFIG_FILES([ scribo/src/binarization/Makefile scribo/src/contest/Makefile scribo/src/contest/hdibco-2010/Makefile + scribo/src/contest/hdlac-2011/Makefile + scribo/src/contest/DAE-2011/Makefile scribo/src/debug/Makefile scribo/src/filter/Makefile scribo/src/misc/Makefile @@ -337,6 +339,7 @@ AC_CONFIG_FILES([ scribo/src/text/Makefile scribo/src/toolchain/Makefile scribo/src/toolchain/nepomuk/Makefile + scribo/src/util/Makefile ]) AC_CONFIG_FILES([scribo/tests/data.hh]) diff --git a/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh b/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh index adfcdb3..5e98f3e 100644 --- a/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh @@ -272,7 +272,8 @@ namespace scribo { on_new_progress_label("Denoise..."); - std::cout << ">> min_area = " << min_area << std::endl; + if (verbose) + std::cout << ">> min_area = " << min_area << std::endl; input_cleaned = preprocessing::denoise_fg(input_cleaned, c8(), min_area); @@ -547,7 +548,8 @@ namespace scribo if (!(closing_size % 2)) closing_size += 1; - std::cout << ">> CLosing size = " << closing_size << std::endl; + if (verbose) + std::cout << ">> CLosing size = " << closing_size << std::endl; component_set<L> elements = scribo::primitive::extract::non_text_hdoc(doc, closing_size); diff --git a/scribo/scribo/toolchain/text_in_doc_preprocess.hh b/scribo/scribo/toolchain/text_in_doc_preprocess.hh index 20fea02..dba5601 100644 --- a/scribo/scribo/toolchain/text_in_doc_preprocess.hh +++ b/scribo/scribo/toolchain/text_in_doc_preprocess.hh @@ -159,6 +159,7 @@ namespace scribo f.sauvola_K = K; f.enable_fg_extraction = enable_fg_bg; f.lambda = lambda; + f.verbose = false; // Get results. mln_ch_value(I,bool) output = f(input); diff --git a/scribo/src/content_in_hdoc.cc b/scribo/src/content_in_hdoc.cc index ed15693..737b1d3 100644 --- a/scribo/src/content_in_hdoc.cc +++ b/scribo/src/content_in_hdoc.cc @@ -31,7 +31,6 @@ #include <mln/core/image/image2d.hh> -#include <mln/io/pbm/save.hh> #include <mln/io/magick/load.hh> #include <scribo/toolchain/content_in_hdoc.hh> @@ -95,8 +94,6 @@ int main(int argc, char* argv[]) scribo::make::internal::debug_filename_prefix = argv[argc - 1]; } - scribo::debug::logger().set_level(scribo::debug::None); - trace::entering("main"); Magick::InitializeMagick(*argv); @@ -198,16 +195,23 @@ int main(int argc, char* argv[]) find_line_seps, find_whitespace_seps, !language.empty(), language); + doc.set_filename(basename(argv[1])); + // Saving results std::cout << "Saving results..." << std::endl; -// scribo::io::xml::save(doc, "page.xml", scribo::io::xml::Page); + scribo::io::xml::save(doc, "page.xml", scribo::io::xml::Page); std::cout << "End of process - " << t << std::endl; - scribo::io::xml::save(doc, argv[2], scribo::io::xml::PageExtended); - scribo::io::img::save(doc, "debug.png", scribo::io::img::DebugWoImage); - scribo::io::img::save(doc, "full.png", scribo::io::img::Full); - // scribo::io::xml::save(doc, "full.xml", scribo::io::xml::Full); +// scribo::io::xml::save(doc, argv[2], scribo::io::xml::PageExtended); + + // scribo::io::img::save(doc, "debug.png", scribo::io::img::DebugWoImage); + // std::cout << "Debug image saved " << t << std::endl; + // scribo::io::img::save(doc, "full.png", scribo::io::img::DebugWithImage); + // std::cout << "Full Debug image saved " << t << std::endl; + // scribo::io::img::save(doc, "full_hd.png", scribo::io::img::Full); + +// sleep(10); trace::exiting("main"); } diff --git a/scribo/tests/estim/Makefile.am b/scribo/src/contest/DAE-2011/Makefile.am similarity index 64% copy from scribo/tests/estim/Makefile.am copy to scribo/src/contest/DAE-2011/Makefile.am index 1c39ab6..09bccd7 100644 --- a/scribo/tests/estim/Makefile.am +++ b/scribo/src/contest/DAE-2011/Makefile.am @@ -17,13 +17,21 @@ ## Process this file through Automake to create Makefile.in. -include $(top_srcdir)/scribo/tests/tests.mk +include $(top_srcdir)/scribo/scribo.mk -check_PROGRAMS = \ - font_color \ - font_boldness +noinst_PROGRAMS = -font_color_SOURCES = font_color.cc -font_boldness_SOURCES = font_boldness.cc -TESTS = $(check_PROGRAMS) +if HAVE_MAGICKXX + +noinst_PROGRAMS += content_in_hdoc_dae + +content_in_hdoc_dae_SOURCES = content_in_hdoc_dae.cc +content_in_hdoc_dae_CPPFLAGS = $(AM_CPPFLAGS) \ + $(TESSERACT_CPPFLAGS) \ + -I$(top_builddir) +content_in_hdoc_dae_LDFLAGS = $(AM_LDFLAGS) \ + -lpthread \ + $(MAGICKXX_LDFLAGS) + +endif HAVE_MAGICKXX \ No newline at end of file diff --git a/scribo/src/non_text_components.cc b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc similarity index 51% copy from scribo/src/non_text_components.cc copy to scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc index 0f4cce4..5309d49 100644 --- a/scribo/src/non_text_components.cc +++ b/scribo/src/contest/DAE-2011/content_in_hdoc_dae.cc @@ -1,4 +1,4 @@ -// Copyright (C) 2011 EPITA Research and Development Laboratory +// Copyright (C) 2010, 2011 EPITA Research and Development Laboratory // (LRDE) // // This file is part of Olena. @@ -24,22 +24,24 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +# define SCRIBO_NOCR #include <libgen.h> #include <fstream> #include <iostream> +#include <sstream> #include <mln/core/image/image2d.hh> -#include <mln/io/pbm/save.hh> #include <mln/io/magick/load.hh> -#include <scribo/toolchain/content_in_doc.hh> +#include <scribo/toolchain/content_in_hdoc.hh> #include <scribo/toolchain/text_in_doc_preprocess.hh> #include <scribo/core/document.hh> #include <scribo/debug/usage.hh> +#include <scribo/debug/logger.hh> #include <scribo/preprocessing/crop_without_localization.hh> #include <scribo/preprocessing/crop.hh> @@ -48,12 +50,16 @@ #include <scribo/io/img/save.hh> +#include <mln/core/alias/neighb2d.hh> +#include <mln/labeling/compute.hh> +#include <mln/labeling/foreground.hh> +#include <mln/util/timer.hh> + + + const char *args_desc[][2] = { - { "input.*", "An image." }, - { "non_text_comps.pbm", "Non text components mask." }, - { "enable_debug", "Enable debug image output. Set to 1 or 0." }, - { "enable_tabstops", "Enable tabstops detection. Set to 1 or 0." }, + { "input.tif", "An image." }, {0, 0} }; @@ -64,65 +70,84 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 4 && argc != 3 && argc != 5) + if (argc != 2) return scribo::debug::usage(argv, - "Extract non text components mask/", - "input.* non_text_comps.pbm [enable_debug] [enable_tabstops]", + "Document Image Analysis in Historical Documents", + "input.tif", args_desc); - std::string out_img = basename(argv[1]); - out_img.erase(out_img.size() - 4); - - std::string filename_prefix = out_img + "_debug"; - scribo::debug::logger().set_filename_prefix(filename_prefix.c_str()); - if (argc > 3 && atoi(argv[3])) - scribo::debug::logger().set_level(scribo::debug::Special); - else - scribo::debug::logger().set_level(scribo::debug::None); - trace::entering("main"); Magick::InitializeMagick(*argv); + mln::util::timer t; + t.start(); + typedef image2d<scribo::def::lbl_type> L; image2d<value::rgb8> input; mln::io::magick::load(input, argv[1]); - util::timer t; - t.start(); - // Preprocess document - image2d<bool> + image2d<bool> input_preproc; + { input_preproc = toolchain::text_in_doc_preprocess(input, false, 0.34); - - bool denoise = true; + // Cleanup components on borders + { + typedef scribo::def::lbl_type V; + V nlabels; + image2d<V> lbl = labeling::foreground(input_preproc, c8(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + + const box2d& b = input.domain(); + for_all_ncomponents(e, nlabels) + if (bbox(e).pmin().row() == b.pmin().row() + || bbox(e).pmax().row() == b.pmax().row() + || bbox(e).pmin().col() == b.pmin().col() + || bbox(e).pmax().col() == b.pmax().col()) + data::fill(((input_preproc | bbox(e)).rw() | (pw::value(lbl) == pw::cst(e))).rw(), false); + } + } + + bool denoise = 1; std::string language = ""; bool find_line_seps = true; - bool find_whitespace_seps = (argc > 4 && atoi(argv[4])); - - std::cout << "Running with the following options :" - << " ocr_language = " << language - << " | find_lines_seps = " << find_line_seps - << " | find_whitespace_seps = " << find_whitespace_seps - << " | debug = " << scribo::debug::logger().is_enabled() - << std::endl; + bool find_whitespace_seps = true; // Run document toolchain. // Text std::cout << "Analysing document..." << std::endl; document<L> - doc = scribo::toolchain::content_in_doc(input, input_preproc, denoise, - find_line_seps, find_whitespace_seps, - !language.empty(), language); - t.stop(); - std::cout << t << std::endl; - - mln::io::pbm::save(data::convert(bool(), doc.elements().labeled_image()), argv[2]); + doc = scribo::toolchain::content_in_hdoc(input, input_preproc, denoise, + find_line_seps, find_whitespace_seps, + !language.empty(), language); + + doc.set_filename(basename(argv[1])); + + const paragraph_set<L>& par_set = doc.paragraphs(); + image2d<bool> output(input.domain()); + for_all_paragraphs(p, par_set) + { + data::fill(output, true); + const paragraph_info<L>& current_par = par_set(p); + const mln::util::array<line_id_t>& line_ids = current_par.line_ids(); + const unsigned nelements = line_ids.nelements(); + + for (unsigned i = 0; i < nelements; ++i) + { + const line_id_t& line_id = line_ids(i); + const line_info<L>& current_line = par_set.lines()(line_id); + + scribo::draw::line_components(output, par_set.lines(), current_line, false); + } + + std::stringstream ss; + ss << basename(argv[1]) << p << ".pbm"; + mln::io::pbm::save(output, ss.str()); + } - scribo::io::img::save(doc, out_img + "_debug_wo_image.png", scribo::io::img::DebugWoImage); - scribo::io::img::save(doc, out_img + "_debug_with_image.png", scribo::io::img::DebugWithImage); trace::exiting("main"); } diff --git a/scribo/src/contest/Makefile.am b/scribo/src/contest/Makefile.am index b1d2dfb..ade408a 100644 --- a/scribo/src/contest/Makefile.am +++ b/scribo/src/contest/Makefile.am @@ -1,4 +1,5 @@ -# Copyright (C) 2010 EPITA Research and Development Laboratory (LRDE). +# Copyright (C) 2010, 2011 EPITA Research and Development Laboratory +# (LRDE). # # This file is part of Olena. # @@ -20,4 +21,6 @@ include $(top_srcdir)/scribo/scribo.mk SUBDIRS = \ - hdibco-2010 \ No newline at end of file + hdibco-2010 \ + hdlac-2011 \ + DAE-2011 \ No newline at end of file diff --git a/scribo/tests/estim/Makefile.am b/scribo/src/contest/hdlac-2011/Makefile.am similarity index 63% copy from scribo/tests/estim/Makefile.am copy to scribo/src/contest/hdlac-2011/Makefile.am index 1c39ab6..f5ead52 100644 --- a/scribo/tests/estim/Makefile.am +++ b/scribo/src/contest/hdlac-2011/Makefile.am @@ -17,13 +17,21 @@ ## Process this file through Automake to create Makefile.in. -include $(top_srcdir)/scribo/tests/tests.mk +include $(top_srcdir)/scribo/scribo.mk -check_PROGRAMS = \ - font_color \ - font_boldness +noinst_PROGRAMS = -font_color_SOURCES = font_color.cc -font_boldness_SOURCES = font_boldness.cc -TESTS = $(check_PROGRAMS) +if HAVE_MAGICKXX + +noinst_PROGRAMS += content_in_hdoc_hdlac + +content_in_hdoc_hdlac_SOURCES = content_in_hdoc_hdlac.cc +content_in_hdoc_hdlac_CPPFLAGS = $(AM_CPPFLAGS) \ + $(TESSERACT_CPPFLAGS) \ + -I$(top_builddir) +content_in_hdoc_hdlac_LDFLAGS = $(AM_LDFLAGS) \ + -lpthread \ + $(MAGICKXX_LDFLAGS) + +endif HAVE_MAGICKXX \ No newline at end of file diff --git a/scribo/src/non_text_components.cc b/scribo/src/contest/hdlac-2011/content_in_hdoc_hdlac.cc similarity index 56% copy from scribo/src/non_text_components.cc copy to scribo/src/contest/hdlac-2011/content_in_hdoc_hdlac.cc index 0f4cce4..045803f 100644 --- a/scribo/src/non_text_components.cc +++ b/scribo/src/contest/hdlac-2011/content_in_hdoc_hdlac.cc @@ -1,4 +1,4 @@ -// Copyright (C) 2011 EPITA Research and Development Laboratory +// Copyright (C) 2010, 2011 EPITA Research and Development Laboratory // (LRDE) // // This file is part of Olena. @@ -24,6 +24,7 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +# define SCRIBO_NOCR #include <libgen.h> #include <fstream> @@ -31,15 +32,15 @@ #include <mln/core/image/image2d.hh> -#include <mln/io/pbm/save.hh> #include <mln/io/magick/load.hh> -#include <scribo/toolchain/content_in_doc.hh> +#include <scribo/toolchain/content_in_hdoc.hh> #include <scribo/toolchain/text_in_doc_preprocess.hh> #include <scribo/core/document.hh> #include <scribo/debug/usage.hh> +#include <scribo/debug/logger.hh> #include <scribo/preprocessing/crop_without_localization.hh> #include <scribo/preprocessing/crop.hh> @@ -48,12 +49,17 @@ #include <scribo/io/img/save.hh> +#include <mln/core/alias/neighb2d.hh> +#include <mln/labeling/compute.hh> +#include <mln/labeling/foreground.hh> +#include <mln/util/timer.hh> + + + const char *args_desc[][2] = { - { "input.*", "An image." }, - { "non_text_comps.pbm", "Non text components mask." }, - { "enable_debug", "Enable debug image output. Set to 1 or 0." }, - { "enable_tabstops", "Enable tabstops detection. Set to 1 or 0." }, + { "input.tif", "An image." }, + { "out.xml", "Result of the document analysis." }, {0, 0} }; @@ -64,65 +70,66 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 4 && argc != 3 && argc != 5) + if (argc != 3) return scribo::debug::usage(argv, - "Extract non text components mask/", - "input.* non_text_comps.pbm [enable_debug] [enable_tabstops]", + "Document Image Analysis in Historical Documents", + "input.tif out.xml", args_desc); - std::string out_img = basename(argv[1]); - out_img.erase(out_img.size() - 4); - - std::string filename_prefix = out_img + "_debug"; - scribo::debug::logger().set_filename_prefix(filename_prefix.c_str()); - if (argc > 3 && atoi(argv[3])) - scribo::debug::logger().set_level(scribo::debug::Special); - else - scribo::debug::logger().set_level(scribo::debug::None); - trace::entering("main"); Magick::InitializeMagick(*argv); + mln::util::timer t; + t.start(); + typedef image2d<scribo::def::lbl_type> L; image2d<value::rgb8> input; mln::io::magick::load(input, argv[1]); - util::timer t; - t.start(); - // Preprocess document - image2d<bool> + image2d<bool> input_preproc; + { input_preproc = toolchain::text_in_doc_preprocess(input, false, 0.34); - - bool denoise = true; + // Cleanup components on borders + { + typedef scribo::def::lbl_type V; + V nlabels; + image2d<V> lbl = labeling::foreground(input_preproc, c8(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + + const box2d& b = input.domain(); + for_all_ncomponents(e, nlabels) + if (bbox(e).pmin().row() == b.pmin().row() + || bbox(e).pmax().row() == b.pmax().row() + || bbox(e).pmin().col() == b.pmin().col() + || bbox(e).pmax().col() == b.pmax().col()) + data::fill(((input_preproc | bbox(e)).rw() | (pw::value(lbl) == pw::cst(e))).rw(), false); + } + } + + bool denoise = 1; std::string language = ""; bool find_line_seps = true; - bool find_whitespace_seps = (argc > 4 && atoi(argv[4])); - - std::cout << "Running with the following options :" - << " ocr_language = " << language - << " | find_lines_seps = " << find_line_seps - << " | find_whitespace_seps = " << find_whitespace_seps - << " | debug = " << scribo::debug::logger().is_enabled() - << std::endl; + bool find_whitespace_seps = true; // Run document toolchain. // Text std::cout << "Analysing document..." << std::endl; document<L> - doc = scribo::toolchain::content_in_doc(input, input_preproc, denoise, - find_line_seps, find_whitespace_seps, - !language.empty(), language); - t.stop(); - std::cout << t << std::endl; + doc = scribo::toolchain::content_in_hdoc(input, input_preproc, denoise, + find_line_seps, find_whitespace_seps, + !language.empty(), language); - mln::io::pbm::save(data::convert(bool(), doc.elements().labeled_image()), argv[2]); + doc.set_filename(basename(argv[1])); - scribo::io::img::save(doc, out_img + "_debug_wo_image.png", scribo::io::img::DebugWoImage); - scribo::io::img::save(doc, out_img + "_debug_with_image.png", scribo::io::img::DebugWithImage); + // Saving results + std::cout << "Saving results..." << std::endl; + scribo::io::xml::save(doc, argv[2], scribo::io::xml::Page); + std::cout << "End of process - " << t << std::endl; trace::exiting("main"); } -- 1.5.6.5

1 0

last-svn-commit-908-gb0db028 Small fixes.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

--- scribo/scribo/io/xml/internal/page_xml_visitor.hh | 15 ++++++++- scribo/scribo/toolchain/content_in_hdoc.hh | 1 + .../toolchain/internal/content_in_hdoc_functor.hh | 32 +++++++++++-------- scribo/scribo/util/component_precise_outline.hh | 4 -- 4 files changed, 32 insertions(+), 20 deletions(-) diff --git a/scribo/scribo/io/xml/internal/page_xml_visitor.hh b/scribo/scribo/io/xml/internal/page_xml_visitor.hh index 8373b02..dab1cce 100644 --- a/scribo/scribo/io/xml/internal/page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/page_xml_visitor.hh @@ -85,6 +85,7 @@ namespace scribo private: // Attributes std::ofstream& output; mutable int base_vertical_line_id_; + mutable int base_text_id_; mutable L lbl_; }; @@ -113,13 +114,23 @@ namespace scribo // 0, so vertical and horizontal lines with the same id // exist. base_vertical_line_id_ = doc.hline_seps_comps().nelements(); + base_text_id_ = 0; // Preambule print_PAGE_preambule(output, doc, true); // Text if (doc.has_text()) + { + + // FIXME: counting the number of valid lines... + for_all_paragraphs(p, doc.paragraphs()) + if (doc.paragraphs()(p).is_valid()) + ++base_text_id_; + --base_text_id_; + doc.paragraphs().accept(*this); + } // Page elements (Pictures, ...) if (doc.has_elements()) @@ -191,8 +202,8 @@ namespace scribo case component::DropCapital: { - output << " <TextRegion id=\"r" << id << "\" " - << " Type=\"Drop_Capital\">" + output << " <TextRegion id=\"r" << base_text_id_ + id << "\" " + << " type=\"drop-capital\">" // FIXME: should not be inline here! << std::endl; internal::print_image_coords(output, par, " "); diff --git a/scribo/scribo/toolchain/content_in_hdoc.hh b/scribo/scribo/toolchain/content_in_hdoc.hh index 97233d5..67d8d2a 100644 --- a/scribo/scribo/toolchain/content_in_hdoc.hh +++ b/scribo/scribo/toolchain/content_in_hdoc.hh @@ -72,6 +72,7 @@ namespace scribo f.enable_whitespace_seps = find_whitespace_seps; f.ocr_language = language; f.enable_ocr = enable_ocr; + f.verbose = false; document<mln_ch_value(I, def::lbl_type)> doc = f(input, input_preproc); diff --git a/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh b/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh index 24d24a3..adfcdb3 100644 --- a/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_hdoc_functor.hh @@ -70,7 +70,10 @@ # include <scribo/postprocessing/images_to_drop_capital.hh> +# ifndef SCRIBO_NOCR # include <scribo/text/recognition.hh> +# endif // ! SCRIBO_NOCR + # include <scribo/text/merging.hh> # include <scribo/text/link_lines.hh> # include <scribo/text/paragraphs.hh> @@ -505,6 +508,8 @@ namespace scribo # endif // ! SCRIBO_NDEBUG //===== END OF DEBUG ===== + +# ifndef SCRIBO_NOCR // Text recognition if (enable_ocr) { @@ -514,6 +519,7 @@ namespace scribo on_progress(); } +# endif // ! SCRIBO_NOCR on_new_progress_label("Extracting paragraphs"); @@ -524,11 +530,11 @@ namespace scribo on_new_progress_label("Filtering paragraphs"); - paragraph_set<L> parset_f = filter::paragraphs_bbox_overlap(parset); - doc.set_paragraphs(parset_f); + // paragraph_set<L> parset_f = filter::paragraphs_bbox_overlap(parset); + // doc.set_paragraphs(parset_f); - // parset = filter::paragraphs_bbox_overlap(parset); - // doc.set_paragraphs(parset); + parset = filter::paragraphs_bbox_overlap(parset); + doc.set_paragraphs(parset); on_progress(); @@ -557,14 +563,14 @@ namespace scribo on_progress(); -// TEMPORARY DEBUG - on_new_progress_label("Saving debug data"); - doc.set_paragraphs(parset); - scribo::io::img::save(doc, "debug_wo_filter.png", scribo::io::img::DebugWoImage); - scribo::io::img::save(doc, "full_wo_filter.png", scribo::io::img::DebugWithImage); - doc.set_paragraphs(parset_f); - on_progress(); -// END OF TEMPORARY DEBUG +// // TEMPORARY DEBUG +// on_new_progress_label("Saving debug data"); +// doc.set_paragraphs(parset); +// scribo::io::img::save(doc, "debug_wo_filter.png", scribo::io::img::DebugWoImage); +// scribo::io::img::save(doc, "full_wo_filter.png", scribo::io::img::DebugWithImage); +// doc.set_paragraphs(parset_f); +// on_progress(); +// // END OF TEMPORARY DEBUG on_new_progress_label("Cleanup miscellaneous false positive"); @@ -604,8 +610,6 @@ namespace scribo on_end(); - sleep(10); - return doc; } diff --git a/scribo/scribo/util/component_precise_outline.hh b/scribo/scribo/util/component_precise_outline.hh index 70fc995..175160c 100644 --- a/scribo/scribo/util/component_precise_outline.hh +++ b/scribo/scribo/util/component_precise_outline.hh @@ -386,13 +386,9 @@ namespace scribo } } - std::cout << "Before filter points - " << points.nsites() << std::endl; - mln::p_array<P> waypoints; internal::filter_points(points, waypoints); - std::cout << "After filter points - " << waypoints.nsites() << std::endl; - trace::exiting("scribo::util::component_precise_outline"); return waypoints; } -- 1.5.6.5

1 0

last-svn-commit-907-ga714708 Fix use of uninitialized values.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

* scribo/primitive/extract/lines_h_thick_and_thin.hh, * scribo/primitive/remove/separators.hh: Initialize border. --- scribo/ChangeLog | 7 +++++++ .../primitive/extract/lines_h_thick_and_thin.hh | 1 + scribo/scribo/primitive/remove/separators.hh | 2 ++ 3 files changed, 10 insertions(+), 0 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index cc7cefc..1814185 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,12 @@ 2011-06-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Fix use of uninitialized values. + + * scribo/primitive/extract/lines_h_thick_and_thin.hh, + * scribo/primitive/remove/separators.hh: Initialize border. + +2011-06-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Fix a serious memory leak. * scribo/core/component_info.hh, diff --git a/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh b/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh index 53a5c32..45b2ad3 100644 --- a/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh +++ b/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh @@ -156,6 +156,7 @@ namespace scribo extension::adjust_fill(input, length / 2, 0); accu::count_value<bool> accu(true); image2d<unsigned> count = accu::transform_line(accu, input, length, 1); + border::fill(count, 0); // FIXME: correct? image2d<value::int_u8> output; initialize(output, count); diff --git a/scribo/scribo/primitive/remove/separators.hh b/scribo/scribo/primitive/remove/separators.hh index 23d5fd1..1b75494 100644 --- a/scribo/scribo/primitive/remove/separators.hh +++ b/scribo/scribo/primitive/remove/separators.hh @@ -88,6 +88,8 @@ namespace scribo mln_concrete(I) output = duplicate(input); border::resize(separators, output.border()); + border::fill(separators, 0); // FIXME: we should rewrite image + // browsing below instead. typedef const mln_value(I)* sep_ptr_t; sep_ptr_t -- 1.5.6.5

1 0

last-svn-commit-915-g3d6ca93 Fix use of uninitialized values.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

* scribo/primitive/extract/lines_h_thick_and_thin.hh, * scribo/primitive/remove/separators.hh: Initialize border. --- scribo/ChangeLog | 7 +++++++ .../primitive/extract/lines_h_thick_and_thin.hh | 1 + scribo/scribo/primitive/remove/separators.hh | 2 ++ 3 files changed, 10 insertions(+), 0 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 16fe8c8..59e3e84 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,12 @@ 2011-06-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Fix use of uninitialized values. + + * scribo/primitive/extract/lines_h_thick_and_thin.hh, + * scribo/primitive/remove/separators.hh: Initialize border. + +2011-06-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Fix a serious memory leak. * scribo/core/component_info.hh, diff --git a/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh b/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh index 53a5c32..45b2ad3 100644 --- a/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh +++ b/scribo/scribo/primitive/extract/lines_h_thick_and_thin.hh @@ -156,6 +156,7 @@ namespace scribo extension::adjust_fill(input, length / 2, 0); accu::count_value<bool> accu(true); image2d<unsigned> count = accu::transform_line(accu, input, length, 1); + border::fill(count, 0); // FIXME: correct? image2d<value::int_u8> output; initialize(output, count); diff --git a/scribo/scribo/primitive/remove/separators.hh b/scribo/scribo/primitive/remove/separators.hh index 23d5fd1..1b75494 100644 --- a/scribo/scribo/primitive/remove/separators.hh +++ b/scribo/scribo/primitive/remove/separators.hh @@ -88,6 +88,8 @@ namespace scribo mln_concrete(I) output = duplicate(input); border::resize(separators, output.border()); + border::fill(separators, 0); // FIXME: we should rewrite image + // browsing below instead. typedef const mln_value(I)* sep_ptr_t; sep_ptr_t -- 1.5.6.5

1 0

last-svn-commit-906-ge6c5288 Fix a serious memory leak.
by Guillaume Lazzara 01 Jun '11

01 Jun '11

* scribo/core/component_info.hh, * scribo/core/component_set.hh, * scribo/core/line_info.hh, * scribo/draw/line_components.hh: Remove holder data. Prevented the containers from being freed. * scribo/io/img/internal/debug_img_visitor.hh, * scribo/io/img/internal/full_img_visitor.hh, * scribo/io/img/save.hh, * scribo/io/xml/internal/page_xml_visitor.hh, * scribo/io/xml/save.hh, * scribo/primitive/extract/lines_h_thick_and_thin.hh, * scribo/primitive/remove/separators.hh, * scribo/text/merging.hh, * scribo/text/paragraphs.hh, * scribo/text/paragraphs_closing.hh, * scribo/toolchain/internal/content_in_hdoc_functor.hh: Update code in order to make holder data useless. --- scribo/ChangeLog | 23 ++++++++++ scribo/scribo/core/component_info.hh | 19 +------- scribo/scribo/core/component_set.hh | 6 +- scribo/scribo/core/line_info.hh | 44 +++++++------------ scribo/scribo/draw/line_components.hh | 7 ++- scribo/scribo/io/img/internal/debug_img_visitor.hh | 37 +++++++++------ scribo/scribo/io/img/internal/full_img_visitor.hh | 34 +++++++++------ scribo/scribo/io/img/save.hh | 6 +- scribo/scribo/io/xml/internal/page_xml_visitor.hh | 28 ++++++------ scribo/scribo/io/xml/save.hh | 2 +- scribo/scribo/text/merging.hh | 46 ++++++++++++------- scribo/scribo/text/paragraphs.hh | 26 ++++++----- scribo/scribo/text/paragraphs_closing.hh | 2 +- 13 files changed, 155 insertions(+), 125 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 6f44931..cc7cefc 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,26 @@ +2011-06-01 Guillaume Lazzara <z(a)lrde.epita.fr> + + Fix a serious memory leak. + + * scribo/core/component_info.hh, + * scribo/core/component_set.hh, + * scribo/core/line_info.hh, + * scribo/draw/line_components.hh: Remove holder data. Prevented + the containers from being freed. + + * scribo/io/img/internal/debug_img_visitor.hh, + * scribo/io/img/internal/full_img_visitor.hh, + * scribo/io/img/save.hh, + * scribo/io/xml/internal/page_xml_visitor.hh, + * scribo/io/xml/save.hh, + * scribo/primitive/extract/lines_h_thick_and_thin.hh, + * scribo/primitive/remove/separators.hh, + * scribo/text/merging.hh, + * scribo/text/paragraphs.hh, + * scribo/text/paragraphs_closing.hh, + * scribo/toolchain/internal/content_in_hdoc_functor.hh: Update + code in order to make holder data useless. + 2011-05-26 Guillaume Lazzara <z(a)lrde.epita.fr> * scribo/util/component_precise_outline.hh: Fix namespace diff --git a/scribo/scribo/core/component_info.hh b/scribo/scribo/core/component_info.hh index 2fa2ad1..b42787f 100644 --- a/scribo/scribo/core/component_info.hh +++ b/scribo/scribo/core/component_info.hh @@ -57,8 +57,7 @@ namespace scribo public: component_info(); - component_info(const component_set<L>& holder, - const component_id_t& id, + component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, unsigned card, @@ -86,8 +85,6 @@ namespace scribo bool is_valid() const; - const component_set<L>& holder() const; - protected: component_id_t id_; mln::box2d bbox_; @@ -98,8 +95,6 @@ namespace scribo component::Tag tag_; component::Type type_; - - component_set<L> holder_; }; @@ -125,14 +120,13 @@ namespace scribo template <typename L> - component_info<L>::component_info(const component_set<L>& holder, - const component_id_t& id, + component_info<L>::component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, unsigned card, component::Type type) : id_(id), bbox_(bbox), mass_center_(mass_center), card_(card), - type_(type), holder_(holder) + type_(type) { if (!bbox.is_valid()) tag_ = component::Ignored; @@ -232,13 +226,6 @@ namespace scribo } - template <typename L> - const component_set<L>& - component_info<L>::holder() const - { - return holder_; - } - template <typename L> std::ostream& diff --git a/scribo/scribo/core/component_set.hh b/scribo/scribo/core/component_set.hh index 4f4cd61..ded64ae 100644 --- a/scribo/scribo/core/component_set.hh +++ b/scribo/scribo/core/component_set.hh @@ -337,7 +337,7 @@ namespace scribo data_->infos_.append(component_info<L>()); // Component 0, i.e. the background. for_all_comp_data(i, attribs) { - component_info<L> info(*this, i, attribs[i].first(), + component_info<L> info(i, attribs[i].first(), attribs[i].second(), attribs[i].second_accu().nsites()); data_->infos_.append(info); } @@ -362,7 +362,7 @@ namespace scribo data_->infos_.append(component_info<L>()); // Component 0, i.e. the background. for_all_comp_data(i, attribs) { - component_info<L> info(*this, i, attribs[i].first(), + component_info<L> info(i, attribs[i].first(), attribs[i].second(), attribs[i].second_accu().nsites(), type); data_->infos_.append(info); @@ -389,7 +389,7 @@ namespace scribo data_->infos_.append(component_info<L>()); // Component 0, i.e. the background. for_all_comp_data(i, attribs) { - component_info<L> info(*this, i, attribs[i].first, + component_info<L> info(i, attribs[i].first, attribs[i].second.first, attribs[i].second.second, type); data_->infos_.append(info); diff --git a/scribo/scribo/core/line_info.hh b/scribo/scribo/core/line_info.hh index 9320416..2913f81 100644 --- a/scribo/scribo/core/line_info.hh +++ b/scribo/scribo/core/line_info.hh @@ -140,13 +140,13 @@ namespace scribo std::string text_; std::string html_text_; - // Line set holding this element. - line_set<L> holder_; - // DEBUG stats< float > meanline_clusters_; stats< float > baseline_clusters_; + component_set<L> components_; + object_links<L> links_; + private: void init_(); }; @@ -282,10 +282,6 @@ namespace scribo /// Force a new computation of statistics. void force_stats_update(); - - /// Returns the line set holding this element. - const line_set<L>& holder() const; - /// Returns the delta used to compute the extended bbox. int delta_of_line() const; @@ -376,7 +372,7 @@ namespace scribo line_info_data<L>::line_info_data(const line_set<L>& holder, const group_info& group) : hidden_(false), tag_(line::None), component_ids_(group.component_ids()), - type_(line::Undefined), holder_(holder) + type_(line::Undefined), components_(holder.components()), links_(holder.links()) { init_(); } @@ -385,7 +381,7 @@ namespace scribo line_info_data<L>::line_info_data(const line_set<L>& holder, const mln::util::array<component_id_t>& component_ids) : hidden_(false), tag_(line::None), component_ids_(component_ids), - type_(line::Undefined), holder_(holder) + type_(line::Undefined), components_(holder.components()), links_(holder.links()) { init_(); } @@ -696,7 +692,7 @@ namespace scribo for_all_elements(i, data_->component_ids_) { unsigned c = data_->component_ids_[i]; - data_->holder_.components_()(c).update_type(type); + data_->components_(c).update_type(type); } } @@ -855,7 +851,7 @@ namespace scribo data_->baseline_ + D, bbox().pmax().col() + delta); - data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->components_.labeled_image().domain()); } @@ -915,7 +911,7 @@ namespace scribo data_->ebbox_.merge(enlarge(b, d_delta)); } - data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->components_.labeled_image().domain()); } else // /other/ IS NOT a text line. { @@ -941,7 +937,7 @@ namespace scribo data_->bbox_.merge(other.bbox()); // Make sure the ebbox is included in the image domain. - data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->components_.labeled_image().domain()); } @@ -975,13 +971,13 @@ namespace scribo // Only for the case of two-character words if (card() == 2) { - const component_set<L>& comp_set = data_->holder_.components(); + const component_set<L>& comp_set = data_->components_; const unsigned c1 = data_->component_ids_(0); const unsigned c2 = data_->component_ids_(1); - if (data_->holder_.components()(c1).type() == component::Punctuation - || data_->holder_.components()(c2).type() == component::Punctuation) + if (data_->components_(c1).type() == component::Punctuation + || data_->components_(c2).type() == component::Punctuation) return false; const mln::box2d& bb1 = comp_set(c1).bbox(); @@ -1026,7 +1022,7 @@ namespace scribo unsigned line_info<L>::get_first_char_height() const { - const component_set<L>& comp_set = data_->holder_.components(); + const component_set<L>& comp_set = data_->components_; const unsigned c1 = data_->components_(0); const mln::box2d& bb1 = comp_set(c1).bbox(); @@ -1120,7 +1116,7 @@ namespace scribo line_info<L>::force_stats_update() { typedef mln_site(L) P; - const component_set<L>& comp_set = data_->holder_.components(); + const component_set<L>& comp_set = data_->components_; // Init. typedef mln::value::int_u<12> median_data_t; @@ -1222,11 +1218,11 @@ namespace scribo // (right link) (left link) // Space between characters. - if (data_->holder_.links()(c) != c) + if (data_->links_(c) != c) { int space = bb.pmin().col() - - comp_set(data_->holder_.links()(c)).bbox().pmax().col() - 1; + - comp_set(data_->links_(c)).bbox().pmax().col() - 1; // -- Ignore overlapped characters. if (space > 0) @@ -1330,14 +1326,6 @@ namespace scribo template <typename L> - const line_set<L>& - line_info<L>::holder() const - { - return data_->holder_; - } - - - template <typename L> std::ostream& operator<<(std::ostream& ostr, const line_info<L>& info) { diff --git a/scribo/scribo/draw/line_components.hh b/scribo/scribo/draw/line_components.hh index 878b2c1..12e7489 100644 --- a/scribo/scribo/draw/line_components.hh +++ b/scribo/scribo/draw/line_components.hh @@ -46,6 +46,7 @@ namespace scribo template <typename L, typename I> void line_components(Image<I>& input_, + const line_set<L>& lines, const line_info<L>& line, const mln_value(I)& value); @@ -56,6 +57,7 @@ namespace scribo template <typename L, typename I> void line_components(Image<I>& input_, + const line_set<L>& lines, const line_info<L>& line, const mln_value(I)& value) { @@ -65,9 +67,8 @@ namespace scribo mln_precondition(input.is_valid()); - const line_set<L>& holder = line.holder(); - const component_set<L>& comp_set = holder.components(); - const L& labeled_image = holder.components().labeled_image(); + const component_set<L>& comp_set = lines.components(); + const L& labeled_image = lines.components().labeled_image(); const mln::util::array<component_id_t>& component_ids = line.component_ids(); diff --git a/scribo/scribo/io/img/internal/debug_img_visitor.hh b/scribo/scribo/io/img/internal/debug_img_visitor.hh index 520a743..cde16ed 100644 --- a/scribo/scribo/io/img/internal/debug_img_visitor.hh +++ b/scribo/scribo/io/img/internal/debug_img_visitor.hh @@ -61,7 +61,8 @@ namespace scribo { - class debug_img_visitor : public doc_serializer<debug_img_visitor> + template <typename L> + class debug_img_visitor : public doc_serializer<debug_img_visitor<L> > { public: // Constructor @@ -69,21 +70,18 @@ namespace scribo unsigned output_ratio); // Visit overloads - template <typename L> void visit(const document<L>& doc) const; - template <typename L> void visit(const component_info<L>& info) const; - template <typename L> void visit(const paragraph_set<L>& parset) const; - template <typename L> void visit(const line_info<L>& line) const; private: // Attributes mln::image2d<value::rgb8>& output; unsigned output_ratio; + mutable L lbl_; private: // Methods box2d compute_bbox(const box2d& b) const; @@ -94,9 +92,9 @@ namespace scribo # ifndef MLN_INCLUDE_ONLY - inline + template <typename L> box2d - debug_img_visitor::compute_bbox(const box2d& b) const + debug_img_visitor<L>::compute_bbox(const box2d& b) const { point2d pmin = b.pmin() / output_ratio, @@ -106,8 +104,8 @@ namespace scribo } - inline - debug_img_visitor::debug_img_visitor(mln::image2d<value::rgb8>& out, + template <typename L> + debug_img_visitor<L>::debug_img_visitor(mln::image2d<value::rgb8>& out, unsigned output_ratio) : output(out), output_ratio(output_ratio) { @@ -119,7 +117,7 @@ namespace scribo // template <typename L> void - debug_img_visitor::visit(const document<L>& doc) const + debug_img_visitor<L>::visit(const document<L>& doc) const { // Text if (doc.has_text()) @@ -129,19 +127,28 @@ namespace scribo if (doc.has_elements()) { for_all_comps(e, doc.elements()) + { + lbl_ = doc.elements().labeled_image(); if (doc.elements()(e).is_valid()) doc.elements()(e).accept(*this); + } } // line seraparators if (doc.has_vline_seps()) + { + lbl_ = doc.vline_seps_comps().labeled_image(); for_all_comps(c, doc.vline_seps_comps()) if (doc.vline_seps_comps()(c).is_valid()) doc.vline_seps_comps()(c).accept(*this); + } if (doc.has_hline_seps()) + { + lbl_ = doc.hline_seps_comps().labeled_image(); for_all_comps(c, doc.hline_seps_comps()) if (doc.hline_seps_comps()(c).is_valid()) doc.hline_seps_comps()(c).accept(*this); + } } @@ -150,13 +157,13 @@ namespace scribo // template <typename L> void - debug_img_visitor::visit(const component_info<L>& info) const + debug_img_visitor<L>::visit(const component_info<L>& info) const { // Getting component outline scribo::def::lbl_type id = (scribo::def::lbl_type)info.id().to_equiv(); - const L& lbl = info.holder().labeled_image(); + //const L& lbl = info.holder().labeled_image(); p_array<point2d> - par = scribo::util::component_precise_outline(lbl | info.bbox(), id); + par = scribo::util::component_precise_outline(lbl_ | info.bbox(), id); switch (info.type()) { @@ -187,7 +194,7 @@ namespace scribo // template <typename L> void - debug_img_visitor::visit(const paragraph_set<L>& parset) const + debug_img_visitor<L>::visit(const paragraph_set<L>& parset) const { const line_set<L>& lines = parset.lines(); @@ -216,7 +223,7 @@ namespace scribo template <typename L> void - debug_img_visitor::visit(const line_info<L>& line) const + debug_img_visitor<L>::visit(const line_info<L>& line) const { point2d pmin = line.bbox().pmin(), diff --git a/scribo/scribo/io/img/internal/full_img_visitor.hh b/scribo/scribo/io/img/internal/full_img_visitor.hh index 7b20970..f31eec1 100644 --- a/scribo/scribo/io/img/internal/full_img_visitor.hh +++ b/scribo/scribo/io/img/internal/full_img_visitor.hh @@ -58,27 +58,26 @@ namespace scribo { - class full_img_visitor : public doc_serializer<full_img_visitor> + template <typename L> + class full_img_visitor : public doc_serializer<full_img_visitor<L> > { public: // Constructor full_img_visitor(mln::image2d<value::rgb8>& out); // Visit overloads - template <typename L> void visit(const document<L>& doc) const; - template <typename L> void visit(const component_info<L>& info) const; - template <typename L> void visit(const paragraph_set<L>& parset) const; - template <typename L> void visit(const line_info<L>& line) const; private: // Attributes mln::image2d<value::rgb8>& output; + + mutable L lbl_; }; @@ -86,8 +85,8 @@ namespace scribo # ifndef MLN_INCLUDE_ONLY - inline - full_img_visitor::full_img_visitor(mln::image2d<value::rgb8>& out) + template <typename L> + full_img_visitor<L>::full_img_visitor(mln::image2d<value::rgb8>& out) : output(out) { mln_assertion(output.is_valid()); @@ -98,7 +97,7 @@ namespace scribo // template <typename L> void - full_img_visitor::visit(const document<L>& doc) const + full_img_visitor<L>::visit(const document<L>& doc) const { // Text if (doc.has_text()) @@ -109,20 +108,29 @@ namespace scribo { const component_set<L>& elts = doc.elements(); for_all_comps(e, elts) + { + lbl_ = elts.labeled_image(); if (elts(e).is_valid()) elts(e).accept(*this); + } } // line seraparators if (doc.has_vline_seps()) + { + lbl_ = doc.vline_seps_comps().labeled_image(); for_all_comps(c, doc.vline_seps_comps()) if (doc.vline_seps_comps()(c).is_valid()) doc.vline_seps_comps()(c).accept(*this); + } if (doc.has_hline_seps()) + { + lbl_ = doc.hline_seps_comps().labeled_image(); for_all_comps(c, doc.hline_seps_comps()) if (doc.hline_seps_comps()(c).is_valid()) doc.hline_seps_comps()(c).accept(*this); + } } @@ -131,13 +139,13 @@ namespace scribo // template <typename L> void - full_img_visitor::visit(const component_info<L>& info) const + full_img_visitor<L>::visit(const component_info<L>& info) const { // Getting component outline scribo::def::lbl_type id = (scribo::def::lbl_type)info.id().to_equiv(); - const L& lbl = info.holder().labeled_image(); + //const L& lbl = info.holder().labeled_image(); p_array<point2d> - par = scribo::util::component_precise_outline(lbl | info.bbox(), id); + par = scribo::util::component_precise_outline(lbl_ | info.bbox(), id); switch (info.type()) { @@ -167,7 +175,7 @@ namespace scribo // template <typename L> void - full_img_visitor::visit(const paragraph_set<L>& parset) const + full_img_visitor<L>::visit(const paragraph_set<L>& parset) const { // const line_set<L>& lines = parset.lines(); @@ -187,7 +195,7 @@ namespace scribo template <typename L> void - full_img_visitor::visit(const line_info<L>& line) const + full_img_visitor<L>::visit(const line_info<L>& line) const { // mln::draw::box(output, line.bbox(), literal::red); diff --git a/scribo/scribo/io/img/save.hh b/scribo/scribo/io/img/save.hh index 04f0a3c..a985d07 100644 --- a/scribo/scribo/io/img/save.hh +++ b/scribo/scribo/io/img/save.hh @@ -150,7 +150,7 @@ namespace scribo { mln_precondition(doc.is_valid()); mln::image2d<value::rgb8> output = duplicate(doc.image()); - scribo::io::img::internal::full_img_visitor f(output); + scribo::io::img::internal::full_img_visitor<L> f(output); doc.accept(f); return output; } @@ -164,7 +164,7 @@ namespace scribo output(box2d(doc.image().domain().pmin() / 4, doc.image().domain().pmax() / 4)); data::fill(output, literal::black); - scribo::io::img::internal::debug_img_visitor f(output, 4); + scribo::io::img::internal::debug_img_visitor<L> f(output, 4); doc.accept(f); return output; } @@ -178,7 +178,7 @@ namespace scribo output = mln::subsampling::antialiased(doc.image(), 4); internal::highlight_mask highlight(0.5f); data::transform_inplace(output, highlight); - scribo::io::img::internal::debug_img_visitor f(output, 4); + scribo::io::img::internal::debug_img_visitor<L> f(output, 4); doc.accept(f); return output; } diff --git a/scribo/scribo/io/xml/internal/page_xml_visitor.hh b/scribo/scribo/io/xml/internal/page_xml_visitor.hh index bbdd3e2..8373b02 100644 --- a/scribo/scribo/io/xml/internal/page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/page_xml_visitor.hh @@ -66,28 +66,27 @@ namespace scribo Its XSD file is located here: http://schema.primaresearch.org/PAGE/gts/pagecontent/2010-03-19/pagecontent… */ - class page_xml_visitor : public doc_serializer<page_xml_visitor> + template <typename L> + class page_xml_visitor : public doc_serializer<page_xml_visitor<L> > { public: // Constructor - page_xml_visitor(std::ofstream& out); + page_xml_visitor<L>(std::ofstream& out); // Visit overloads - template <typename L> void visit(const document<L>& doc) const; - template <typename L> void visit(const component_set<L>& comp_set) const; - template <typename L> void visit(const component_info<L>& info) const; - template <typename L> void visit(const paragraph_set<L>& parset) const; private: // Attributes std::ofstream& output; mutable int base_vertical_line_id_; + + mutable L lbl_; }; @@ -95,8 +94,8 @@ namespace scribo # ifndef MLN_INCLUDE_ONLY - inline - page_xml_visitor::page_xml_visitor(std::ofstream& out) + template <typename L> + page_xml_visitor<L>::page_xml_visitor(std::ofstream& out) : output(out) { } @@ -107,7 +106,7 @@ namespace scribo // template <typename L> void - page_xml_visitor::visit(const document<L>& doc) const + page_xml_visitor<L>::visit(const document<L>& doc) const { // Make sure there are no duplicate ids for line separators. // Vertical and horizontal lines are indexed separately from @@ -143,8 +142,9 @@ namespace scribo // template <typename L> void - page_xml_visitor::visit(const component_set<L>& comp_set) const + page_xml_visitor<L>::visit(const component_set<L>& comp_set) const { + lbl_ = comp_set.labeled_image(); for_all_comps(c, comp_set) if (comp_set(c).is_valid()) comp_set(c).accept(*this); @@ -155,13 +155,13 @@ namespace scribo // template <typename L> void - page_xml_visitor::visit(const component_info<L>& info) const + page_xml_visitor<L>::visit(const component_info<L>& info) const { // Getting component outline scribo::def::lbl_type id = (scribo::def::lbl_type)info.id().to_equiv(); - const L& lbl = info.holder().labeled_image(); + //const L& lbl = info.holder().labeled_image(); p_array<point2d> - par = scribo::util::component_precise_outline(lbl | info.bbox(), id); + par = scribo::util::component_precise_outline(lbl_ | info.bbox(), id); switch (info.type()) { @@ -223,7 +223,7 @@ namespace scribo // template <typename L> void - page_xml_visitor::visit(const paragraph_set<L>& parset) const + page_xml_visitor<L>::visit(const paragraph_set<L>& parset) const { const line_set<L>& lines = parset.lines(); diff --git a/scribo/scribo/io/xml/save.hh b/scribo/scribo/io/xml/save.hh index 54afa79..cc6905b 100644 --- a/scribo/scribo/io/xml/save.hh +++ b/scribo/scribo/io/xml/save.hh @@ -96,7 +96,7 @@ namespace scribo template <typename L> void save_page(const document<L>& doc, std::ofstream& output) { - scribo::io::xml::internal::page_xml_visitor f(output); + scribo::io::xml::internal::page_xml_visitor<L> f(output); doc.accept(f); } diff --git a/scribo/scribo/text/merging.hh b/scribo/scribo/text/merging.hh index f1135ed..31a5ed4 100644 --- a/scribo/scribo/text/merging.hh +++ b/scribo/scribo/text/merging.hh @@ -224,11 +224,15 @@ namespace scribo template <typename L> - bool between_separators(const scribo::line_info<L>& l1, - const scribo::line_info<L>& l2) + bool between_separators(const scribo::line_set<L>& lines, + const line_id_t& l1_, + const line_id_t& l2_) { + const scribo::line_info<L>& l1 = lines(l1_); + const scribo::line_info<L>& l2 = lines(l2_); + // No separators found in image. - mln_precondition(l1.holder().components().has_separators()); + mln_precondition(lines.components().has_separators()); const box2d& l1_bbox = l1.bbox(); const box2d& l2_bbox = l2.bbox(); @@ -237,7 +241,7 @@ namespace scribo col1 = l1_bbox.pcenter().col(), col2 = l2_bbox.pcenter().col(); const mln_ch_value(L, bool)& - separators = l1.holder().components().separators(); + separators = lines.components().separators(); // Checking for separators starting from 1 / 4, 3/ 4 and the // center of the box @@ -290,9 +294,13 @@ namespace scribo */ template <typename L> - bool lines_can_merge(scribo::line_info<L>& l1, - const scribo::line_info<L>& l2) + bool lines_can_merge(scribo::line_set<L>& lines, + const scribo::line_id_t& l1_, + const scribo::line_id_t& l2_) { + scribo::line_info<L>& l1 = lines(l1_); + scribo::line_info<L>& l2 = lines(l2_); + // Parameters. const float x_ratio_max = 1.7f; const float baseline_delta_max = @@ -306,9 +314,9 @@ namespace scribo const point2d& l1_pmax = l1_bbox.pmax(); const point2d& l2_pmax = l2_bbox.pmax(); - const bool l1_has_separators = l1.holder().components().has_separators(); + const bool l1_has_separators = lines.components().has_separators(); const bool l1_l2_between_separators = (l1_has_separators) ? - between_separators(l1, l2) : false; + between_separators(lines, l1_, l2_) : false; const float l_ted_cw = l2.char_width(); const float dx = std::max(l1_pmin.col(), l2_pmin.col()) @@ -424,9 +432,13 @@ namespace scribo */ template <typename L> - bool non_text_and_text_can_merge(scribo::line_info<L>& l_cur, // current - const scribo::line_info<L>& l_ted) // touched + bool non_text_and_text_can_merge(scribo::line_set<L>& lines, + const scribo::line_id_t& l_cur_, // current + const scribo::line_id_t l_ted_) // touched { + scribo::line_info<L>& l_cur = lines(l_cur_); + scribo::line_info<L>& l_ted = lines(l_ted_); + if (l_cur.type() == line::Text || l_ted.type() != line::Text) return false; // the current object is a NON-textline @@ -434,8 +446,8 @@ namespace scribo // Check that there is no separator in between. - if (l_cur.holder().components().has_separators() - && between_separators(l_cur, l_ted)) + if (lines.components().has_separators() + && between_separators(lines, l_cur_, l_ted_)) return false; const box2d& l_cur_bbox = l_cur.bbox(); @@ -744,8 +756,8 @@ namespace scribo < 5 && std::abs(l_info.meanline() - mc_info.meanline()) < 5)) && dx < l_ted_cw && dy < 0 - && not (l_info.holder().components().has_separators() - && between_separators(l_info, mc_info))) + && not (lines.components().has_separators() + && between_separators(lines, l, mc))) l = do_union(lines, l, mc, parent); // } @@ -801,7 +813,7 @@ namespace scribo // could be noise or garbage... So adding new // criterions could fix this issue. // - if (!non_text_and_text_can_merge(lines(l), lines(mc))) + if (!non_text_and_text_can_merge(lines, l, mc)) continue; // Avoid the case when a large title ebbox overlap @@ -868,7 +880,7 @@ namespace scribo if (lines(l_).type() == line::Text) { // l_ and lcand look like text line chunks. - if (lines_can_merge(lines(l_), lines(lcand))) + if (lines_can_merge(lines, l_, lcand)) { ++count_two_lines_merge; l_ = do_union(lines, l_, lcand, parent); @@ -899,7 +911,7 @@ namespace scribo { // l_ does NOT looks like a text line chunk. ++count_comp_HITS_txtline; - if (non_text_and_text_can_merge(lines(l_), lines(lcand))) + if (non_text_and_text_can_merge(lines, l_, lcand)) // a petouille merges with a text line? { ++count_comp_HITS_txtline; diff --git a/scribo/scribo/text/paragraphs.hh b/scribo/scribo/text/paragraphs.hh index 8fd89be..e37f610 100644 --- a/scribo/scribo/text/paragraphs.hh +++ b/scribo/scribo/text/paragraphs.hh @@ -29,11 +29,15 @@ namespace scribo template <typename L> inline bool - between_horizontal_separator(const scribo::line_info<L>& l1, - const scribo::line_info<L>& l2) + between_horizontal_separator(const line_set<L>& lines, + const line_id_t& l1_, + const line_id_t& l2_) { + const line_info<L>& l1 = lines(l1_); + const line_info<L>& l2 = lines(l2_); + // No separators found in image. - mln_precondition(l1.holder().components().has_separators()); + mln_precondition(lines.components().has_separators()); const box2d& l1_bbox = l1.bbox(); const box2d& l2_bbox = l2.bbox(); @@ -42,7 +46,7 @@ namespace scribo row1 = l1_bbox.pcenter().row(), row2 = l2_bbox.pcenter().row(); const mln_ch_value(L, bool)& - separators = l1.holder().components().separators(); + separators = lines.components().separators(); unsigned row; unsigned col_ptr; @@ -153,18 +157,18 @@ namespace scribo line_id_t right_nbh = right(l); line_id_t lol_nbh = output(left_nbh); - const line_info<L>& left_line = lines(left_nbh); - const line_info<L>& current_line = lines(l); - const line_info<L>& right_line = lines(right_nbh); + // const line_info<L>& left_line = lines(left_nbh); + // const line_info<L>& current_line = lines(l); + // const line_info<L>& right_line = lines(right_nbh); - if (right_line.holder().components().has_separators() && - between_horizontal_separator(right_line, current_line)) + if (lines.components().has_separators() && + between_horizontal_separator(lines, right_nbh, l)) { output(right_nbh) = right_nbh; right_nbh = l; } - if (current_line.holder().components().has_separators() && - between_horizontal_separator(current_line, left_line)) + if (lines.components().has_separators() && + between_horizontal_separator(lines, l, left_nbh)) { output(l) = l; left_nbh = l; diff --git a/scribo/scribo/text/paragraphs_closing.hh b/scribo/scribo/text/paragraphs_closing.hh index efc5259..2b685df 100644 --- a/scribo/scribo/text/paragraphs_closing.hh +++ b/scribo/scribo/text/paragraphs_closing.hh @@ -210,7 +210,7 @@ namespace scribo const line_id_t& line_id = line_ids(i); const line_info<L>& current_line = lines(line_id); - scribo::draw::line_components(debug, current_line, p); + scribo::draw::line_components(debug, lines, current_line, p); // HACK DISCLAIMER : this line is drawn in order to be // sure that every line will be reduced to a single -- 1.5.6.5

1 0