March 2011 - Olena-patches

last-svn-commit-783-g38639f4 Add new debug tools.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

* scribo/src/debug/Makefile.am: New targets. * scribo/src/debug/show_separators.cc, * scribo/src/debug/show_stoppers.cc, * scribo/src/debug/show_text_lines.cc: New. --- scribo/ChangeLog | 10 ++ scribo/src/debug/Makefile.am | 12 ++- .../lines_pattern.cc => debug/show_separators.cc} | 74 +++++++++----- scribo/src/debug/show_stoppers.cc | 104 ++++++++++++++++++++ .../show_text_lines.cc} | 100 +++++++++++-------- 5 files changed, 225 insertions(+), 75 deletions(-) copy scribo/src/{primitive/extract/lines_pattern.cc => debug/show_separators.cc} (50%) create mode 100644 scribo/src/debug/show_stoppers.cc copy scribo/src/{text/pbm_lines_recognition.cc => debug/show_text_lines.cc} (52%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 74c7a03..4a9dcbf 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,15 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new debug tools. + + * scribo/src/debug/Makefile.am: New targets. + + * scribo/src/debug/show_separators.cc, + * scribo/src/debug/show_stoppers.cc, + * scribo/src/debug/show_text_lines.cc: New. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/text/extract_lines.hh: Update code. 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> diff --git a/scribo/src/debug/Makefile.am b/scribo/src/debug/Makefile.am index d203f5c..cdb1f30 100644 --- a/scribo/src/debug/Makefile.am +++ b/scribo/src/debug/Makefile.am @@ -1,5 +1,5 @@ -# Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -# (LRDE). +# Copyright (C) 2009, 2010, 2011 EPITA Research and Development +# Laboratory (LRDE). # # This file is part of Olena. # @@ -20,6 +20,7 @@ include $(top_srcdir)/scribo/scribo.mk noinst_PROGRAMS = \ show_info_x_height \ show_info_median_inter_characters \ + show_separators \ show_links_left_right_links_validation \ show_links_bbox_h_ratio \ show_links_bbox_overlap \ @@ -43,11 +44,14 @@ noinst_PROGRAMS = \ show_objects_large_small \ show_objects_small \ show_objects_thick \ - show_objects_thin + show_objects_thin \ + show_stoppers \ + show_text_lines show_info_x_height_SOURCES = show_info_x_height.cc show_info_median_inter_characters_SOURCES = show_info_median_inter_characters.cc +show_separators_SOURCES = show_separators.cc show_links_left_right_links_validation_SOURCES = show_links_left_right_links_validation.cc show_links_bbox_h_ratio_SOURCES = show_links_bbox_h_ratio.cc show_links_bbox_overlap_SOURCES = show_links_bbox_overlap.cc @@ -72,6 +76,8 @@ show_objects_large_small_SOURCES = show_objects_large_small.cc show_objects_small_SOURCES = show_objects_small.cc show_objects_thick_SOURCES = show_objects_thick.cc show_objects_thin_SOURCES = show_objects_thin.cc +show_stoppers_SOURCES = show_stoppers.cc +show_text_lines_SOURCES = show_text_lines.cc if HAVE_MAGICKXX diff --git a/scribo/src/primitive/extract/lines_pattern.cc b/scribo/src/debug/show_separators.cc similarity index 50% copy from scribo/src/primitive/extract/lines_pattern.cc copy to scribo/src/debug/show_separators.cc index cb56b36..ef80da6 100644 --- a/scribo/src/primitive/extract/lines_pattern.cc +++ b/scribo/src/debug/show_separators.cc @@ -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. // @@ -14,33 +13,25 @@ // // You should have received a copy of the GNU General Public License // along with Olena. If not, see <http://www.gnu.org/licenses/>. -// -// As a special exception, you may use this file as part of a free -// software project without restriction. Specifically, if other files -// instantiate templates or use macros or inline functions from this -// file, or you compile this file and link it with other files to produce -// an executable, this file does not by itself cause the resulting -// executable to be covered by the GNU General Public License. This -// exception does not however invalidate any other reasons why the -// executable file might be covered by the GNU General Public License. #include <mln/core/image/image2d.hh> +#include <mln/core/alias/box2d.hh> +#include <mln/core/alias/neighb2d.hh> #include <mln/io/pbm/all.hh> -#include <mln/pw/all.hh> -#include <mln/data/fill.hh> -#include <mln/core/image/dmorph/image_if.hh> +#include <mln/arith/plus.hh> +#include <mln/labeling/foreground.hh> #include <scribo/primitive/extract/lines_h_pattern.hh> #include <scribo/primitive/extract/lines_v_pattern.hh> #include <scribo/debug/usage.hh> +#include <scribo/debug/save_bboxes_image.hh> const char *args_desc[][2] = { - { "input.pbm", "A binary image." }, - { "length", " Minimum line length. (Common value : 51)" }, + { "length", " Minimum line length. (default: 101)" }, { "delta", " Distance between the object pixel and the background pixel" - "(Common value : 4)" }, + "(default: 4)" }, {0, 0} }; @@ -49,10 +40,10 @@ int main(int argc, char *argv[]) { using namespace mln; - if (argc != 5) + if (argc != 7 && argc != 5) return scribo::debug::usage(argv, - "Extract horizontal lines patterns", - "input.pbm length delta output.pbm", + "Extract horizontal and vertical lines patterns", + "input.pbm output.pbm input_wo_seps.pbm output.ppm length delta", args_desc); trace::entering("main"); @@ -60,19 +51,46 @@ int main(int argc, char *argv[]) image2d<bool> input; io::pbm::load(input, argv[1]); + // Set default option value + unsigned + length = 101, + delta = 4; + if (argc > 5) + { + length = atoi(argv[5]); + delta = atoi(argv[6]); + } + + util::timer t; + t.start(); + image2d<bool> - h_lines = scribo::primitive::extract::lines_h_pattern(input, - atoi(argv[2]), - atoi(argv[3])); + h_lines = scribo::primitive::extract::lines_h_pattern(input, length, delta); image2d<bool> - v_lines = scribo::primitive::extract::lines_v_pattern(input, - atoi(argv[2]), - atoi(argv[3])); + v_lines = scribo::primitive::extract::lines_v_pattern(input, length, delta); + + v_lines += h_lines; + + t.stop(); + std::cout << t << "s" << std::endl; + + // Save binary image. + io::pbm::save(v_lines, argv[2]); - data::fill((v_lines | pw::value(h_lines)).rw(), true); + // Save input without separators + { + image2d<bool> in_wo_seps = duplicate(input); + data::fill((in_wo_seps | pw::value(v_lines)).rw(), false); + io::pbm::save(in_wo_seps, argv[3]); + } - io::pbm::save(v_lines, argv[4]); + // Save bbox image + value::int_u8 nlabels; + image2d<value::int_u8> lbl = labeling::foreground(v_lines, c4(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + scribo::debug::save_bboxes_image(input, bbox, argv[4], literal::red); trace::exiting("main"); } diff --git a/scribo/src/debug/show_stoppers.cc b/scribo/src/debug/show_stoppers.cc new file mode 100644 index 0000000..a108da1 --- /dev/null +++ b/scribo/src/debug/show_stoppers.cc @@ -0,0 +1,104 @@ +// 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/>. + +#include <mln/core/image/image2d.hh> +#include <mln/core/alias/box2d.hh> +#include <mln/core/alias/neighb2d.hh> +#include <mln/io/pbm/all.hh> +#include <mln/arith/plus.hh> +#include <mln/labeling/foreground.hh> + +#include <scribo/primitive/extract/lines_h_pattern.hh> +#include <scribo/primitive/extract/lines_v_pattern.hh> +#include <scribo/primitive/extract/separators_nonvisible.hh> +#include <scribo/debug/usage.hh> +#include <scribo/debug/save_bboxes_image.hh> + + +const char *args_desc[][2] = +{ + { "length", " Minimum line length. (default: 101)" }, + { "delta", " Distance between the object pixel and the background pixel" + "(default: 4)" }, + {0, 0} +}; + + +int main(int argc, char *argv[]) +{ + using namespace mln; + + if (argc != 7 && argc != 5) + return scribo::debug::usage(argv, + "Extract horizontal, vertical lines and stoppers", + "input.pbm output.pbm input_wo_seps.pbm output.ppm length delta", + args_desc); + + trace::entering("main"); + + image2d<bool> input; + io::pbm::load(input, argv[1]); + + // Set default option value + unsigned + length = 101, + delta = 4; + if (argc > 5) + { + length = atoi(argv[5]); + delta = atoi(argv[6]); + } + + util::timer t; + + t.start(); + image2d<bool> + h_lines = scribo::primitive::extract::lines_h_pattern(input, length, delta); + + image2d<bool> + v_lines = scribo::primitive::extract::lines_v_pattern(input, length, delta); + + v_lines += h_lines; + + t.stop(); + + // Save input without separators + { + image2d<bool> in_wo_seps = duplicate(input); + data::fill((in_wo_seps | pw::value(v_lines)).rw(), false); + io::pbm::save(in_wo_seps, argv[3]); + } + + // Save bbox image + value::int_u8 nlabels; + image2d<value::int_u8> lbl = labeling::foreground(v_lines, c4(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + scribo::debug::save_bboxes_image(input, bbox, argv[4], literal::red); + + t.resume(); + + // Non visible separators + v_lines += scribo::primitive::extract::separators_nonvisible(input); + + t.stop(); + std::cout << t << "s" << std::endl; + + // Save binary image. + io::pbm::save(v_lines, argv[2]); + + trace::exiting("main"); +} diff --git a/scribo/src/text/pbm_lines_recognition.cc b/scribo/src/debug/show_text_lines.cc similarity index 52% copy from scribo/src/text/pbm_lines_recognition.cc copy to scribo/src/debug/show_text_lines.cc index 3451449..a031072 100644 --- a/scribo/src/text/pbm_lines_recognition.cc +++ b/scribo/src/debug/show_text_lines.cc @@ -1,4 +1,4 @@ -// Copyright (C) 2010 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -23,15 +23,15 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -#ifdef HAVE_CONFIG_H -# include <config.h> -#endif - #include <iostream> #include <mln/core/image/image2d.hh> #include <mln/io/pbm/load.hh> #include <mln/value/int_u16.hh> +#include <mln/draw/box_plain.hh> +#include <mln/pw/all.hh> +#include <mln/core/image/dmorph/image_if.hh> +#include <mln/data/convert.hh> #include <scribo/text/recognition.hh> @@ -41,19 +41,15 @@ #include <scribo/core/object_links.hh> #include <scribo/core/object_groups.hh> -#include <scribo/text/merging.hh> -#include <scribo/primitive/extract/components.hh> -#include <scribo/primitive/link/with_single_left_link_dmax_ratio.hh> -#include <scribo/primitive/group/from_single_link.hh> +#include <scribo/text/extract_lines.hh> #include <scribo/io/text_boxes/save.hh> +#include <scribo/debug/save_bboxes_image.hh> + const char *args_desc[][2] = { - { "input.pbm", "A binary image. 'True' for objects, 'False'\ -for the background." }, - { "lang", "Must be set to \"eng\", \"fra\", \"deu\", \"spa\", \"ita\" (Default \"fra\")" }, {0, 0} }; @@ -64,10 +60,10 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 2 && argc != 3 && argc != 4) + if (argc != 6) return scribo::debug::usage(argv, - "Text recognition", - "input.pbm [lang] [file]", + "Show text lines", + "input.pbm input_seps.pbm out.ppm out.pbm comps.pbm", args_desc); trace::entering("main"); @@ -75,37 +71,53 @@ int main(int argc, char* argv[]) image2d<bool> input; mln::io::pbm::load(input, argv[1]); - typedef image2d<value::int_u16> L; - value::int_u16 ncomps; - component_set<L> - comps = scribo::primitive::extract::components(input, c8(), ncomps); - std::cout << "ncomps = " << ncomps << std::endl; - - object_links<L> - links = scribo::primitive::link::with_single_left_link_dmax_ratio(comps, 2); - - object_groups<L> - groups = scribo::primitive::group::from_single_link(links); + image2d<bool> input_seps; + mln::io::pbm::load(input_seps, argv[2]); - line_set<L> lines(groups); - lines = text::merging(lines); + util::timer t; + t.start(); - std::string str = argv[2]; - const char *lang; - if (argc < 3 || (str != "eng" && str != "fra" - && str != "deu" && str != "spa" - && str != "ita")) - lang = "fra"; - else - lang = argv[2]; - - char *output = 0; - if (argc >= 4) - output = argv[3]; - - scribo::text::recognition(lines, lang); - - scribo::io::text_boxes::save(lines, output); + typedef image2d<value::int_u16> L; + line_set<L> lines = text::extract_lines(input, c8(), input_seps); + + t.stop(); + std::cout << t << "s" << std::endl; + + // Color image with line bboxes. + { + image2d<value::rgb8> output = data::convert(value::rgb8(), input); + for_all_lines(l, lines) + if (lines(l).is_textline()) + mln::draw::box(output, lines(l).bbox(), literal::red); + mln::io::ppm::save(output, argv[3]); + } + + // Binary image with line bboxes. + { + image2d<bool> output; + initialize(output, input); + data::fill(output, false); + for_all_lines(l, lines) + if (lines(l).is_textline()) + mln::draw::box_plain(output, lines(l).bbox(), true); + mln::io::pbm::save(output, argv[4]); + } + + { + image2d<bool> output; + initialize(output, input); + data::fill(output, false); + for_all_lines(l, lines) + if (lines(l).is_textline()) + { + const util::array<component_id_t>& comps = lines(l).components(); + const L& lbl = lines.components().labeled_image(); + for_all_elements(c, comps) + data::fill((output | lines.components()(comps(c)).bbox()).rw(), + ((input | lines.components()(comps(c)).bbox()) | (pw::value(lbl) == comps(c)))); + } + mln::io::pbm::save(output, argv[5]); + } trace::exiting("main"); } -- 1.5.6.5

1 0

last-svn-commit-782-g08507d9 scribo/text/extract_lines.hh: Update code.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

--- scribo/ChangeLog | 4 ++ scribo/scribo/text/extract_lines.hh | 68 ++++++++++++++++++++++------------ 2 files changed, 48 insertions(+), 24 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index f740230..74c7a03 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/text/extract_lines.hh: Update code. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make use of mln::fun::v2v::rgb_to_luma. * scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, diff --git a/scribo/scribo/text/extract_lines.hh b/scribo/scribo/text/extract_lines.hh index 1a25cc8..b81cb79 100644 --- a/scribo/scribo/text/extract_lines.hh +++ b/scribo/scribo/text/extract_lines.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -33,13 +33,14 @@ # include <mln/core/concept/image.hh> -# include <mln/value/label_16.hh> +# include <mln/value/int_u16.hh> # include <scribo/core/line_set.hh> # include <scribo/primitive/extract/components.hh> # include <scribo/primitive/link/merge_double_link.hh> +# include <scribo/primitive/link/internal/dmax_width_and_height.hh> # include <scribo/primitive/link/with_single_left_link_dmax_ratio.hh> # include <scribo/primitive/link/with_single_right_link_dmax_ratio.hh> @@ -65,11 +66,18 @@ namespace scribo /*! ** \param[in] input A binary image. ** \param[in] nbh A neighborhood used for labeling. + ** \param[in] seps A binary image with separator information. ** ** \return A set of lines. */ template <typename I, typename N> - line_set<mln_ch_value(I,value::label_16)> + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input_, const Neighborhood<N>& nbh_, + const mln_ch_value(I,bool)& separators); + + /// \overload + template <typename I, typename N> + line_set<mln_ch_value(I,value::int_u16)> extract_lines(const Image& input, const Neighborhood<N>& nbh); @@ -77,8 +85,18 @@ namespace scribo template <typename I, typename N> - line_set<mln_ch_value(I,value::label_16)> - extract_lines(const Image& input_, const Neighborhood<N>& nbh_) + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input, const Neighborhood<N>& nbh) + { + mln_ch_value(I,bool) seps; + return extract_lines(input, nbh, seps); + } + + + template <typename I, typename N> + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input_, const Neighborhood<N>& nbh_, + const mln_ch_value(I,bool)& separators) { trace::entering("scribo::text::extract_lines"); @@ -88,44 +106,46 @@ namespace scribo mln_precondition(input.is_valid()); mln_precondition(nbh.is_valid()); - typedef mln_ch_value(I,value::label_16) L; - /// Finding comps. - value::label_16 ncomps; + typedef mln_ch_value(I,value::int_u16) L; + value::int_u16 ncomps; component_set<L> - comps = scribo::primitive::extract::components(input, nbh, ncomps); + comps = scribo::primitive::extract::components(input, c8(), ncomps); /// First filtering. - component_set<L> filtered_comps - = scribo::filter::components_small(comps, 6); + comps = scribo::filter::components_small(comps, 3); + + /// Use separators. + if (exact(separators).is_valid()) + comps.add_separators(separators); /// Linking potential comps object_links<L> left_link - = primitive::link::with_single_left_link_dmax_ratio(filtered_comps); + = primitive::link::with_single_left_link_dmax_ratio(comps, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); object_links<L> right_link - = primitive::link::with_single_right_link_dmax_ratio(filtered_comps); - + = primitive::link::with_single_right_link_dmax_ratio(comps, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); // Validating left and right links. object_links<L> merged_links = primitive::link::merge_double_link(left_link, - right_link); + right_link); - // Remove links if bboxes have too different sizes. - object_links<L> hratio_filtered_links - = filter::object_links_bbox_h_ratio(merged_links, 2.0f); + object_links<L> hratio_filtered_links + = filter::object_links_bbox_h_ratio(merged_links, 2.5f); object_groups<L> groups = primitive::group::from_single_link(hratio_filtered_links); - - line_set<L> line = scribo::make::line_set(groups); - line = text::merging(line); - + line_set<L> lines(groups); + lines = text::merging(lines); trace::exiting("scribo::text::extract_lines"); - return line; + return lines; } # endif // ! MLN_INCLUDE_ONLY -- 1.5.6.5

1 0

last-svn-commit-781-g300d0d7 Make use of mln::fun::v2v::rgb_to_luma.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

* scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, * src/binarization/ppm_sauvola.cc, * src/binarization/ppm_sauvola_ms.cc, * src/binarization/ppm_sauvola_ms_fg.cc, * src/binarization/ppm_sauvola_ms_split.cc, * src/binarization/sauvola.cc, * src/binarization/sauvola_debug.cc, * src/binarization/sauvola_ms.cc, * src/binarization/sauvola_ms_debug.cc, * src/binarization/sauvola_ms_fg.cc, * src/text_in_picture.cc, * src/text_in_picture_neg.cc, * src/text_recognition_in_picture.cc: Here. --- scribo/ChangeLog | 18 ++++++++++++++++++ .../internal/text_in_doc_preprocess_functor.hh | 7 ++++--- scribo/src/binarization/ppm_sauvola.cc | 9 +++++---- scribo/src/binarization/ppm_sauvola_ms.cc | 9 +++++---- scribo/src/binarization/ppm_sauvola_ms_fg.cc | 8 ++++---- scribo/src/binarization/ppm_sauvola_ms_split.cc | 5 ++--- scribo/src/binarization/sauvola.cc | 4 ++-- scribo/src/binarization/sauvola_debug.cc | 8 ++++---- scribo/src/binarization/sauvola_ms.cc | 9 +++++---- scribo/src/binarization/sauvola_ms_debug.cc | 9 +++++---- scribo/src/binarization/sauvola_ms_fg.cc | 8 ++++---- scribo/src/text_in_picture.cc | 7 ++++--- scribo/src/text_in_picture_neg.cc | 5 ++--- scribo/src/text_recognition_in_picture.cc | 9 +++++---- 14 files changed, 69 insertions(+), 46 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index cf51ca8..f740230 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,23 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make use of mln::fun::v2v::rgb_to_luma. + + * scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, + * src/binarization/ppm_sauvola.cc, + * src/binarization/ppm_sauvola_ms.cc, + * src/binarization/ppm_sauvola_ms_fg.cc, + * src/binarization/ppm_sauvola_ms_split.cc, + * src/binarization/sauvola.cc, + * src/binarization/sauvola_debug.cc, + * src/binarization/sauvola_ms.cc, + * src/binarization/sauvola_ms_debug.cc, + * src/binarization/sauvola_ms_fg.cc, + * src/text_in_picture.cc, + * src/text_in_picture_neg.cc, + * src/text_recognition_in_picture.cc: Here. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/io/xml/load.hh: New XML loader. 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> diff --git a/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh b/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh index 6a9506b..6c0dd5a 100644 --- a/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh +++ b/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh @@ -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. // @@ -29,7 +30,7 @@ #include <mln/core/concept/image.hh> #include <mln/data/transform.hh> #include <mln/data/convert.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/subsampling/antialiased.hh> #include <scribo/binarization/sauvola.hh> @@ -208,7 +209,7 @@ namespace scribo on_new_progress_label("Convert to gray-scale image"); image2d<value::int_u8> intensity_ima = mln::data::transform(input_rgb, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); on_progress(); diff --git a/scribo/src/binarization/ppm_sauvola.cc b/scribo/src/binarization/ppm_sauvola.cc index f0cd355..4732a02 100644 --- a/scribo/src/binarization/ppm_sauvola.cc +++ b/scribo/src/binarization/ppm_sauvola.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -27,7 +27,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola.hh> #include <scribo/debug/usage.hh> @@ -74,7 +74,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_gl = data::transform(input, mln::fun::v2v::rgb_to_int_u<8>()); + input_gl = data::transform(input, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> out = scribo::binarization::sauvola(input_gl, w, k); diff --git a/scribo/src/binarization/ppm_sauvola_ms.cc b/scribo/src/binarization/ppm_sauvola_ms.cc index eb694c2..cff28f9 100644 --- a/scribo/src/binarization/ppm_sauvola_ms.cc +++ b/scribo/src/binarization/ppm_sauvola_ms.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -29,7 +29,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -110,7 +110,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/ppm_sauvola_ms_fg.cc b/scribo/src/binarization/ppm_sauvola_ms_fg.cc index 20f237f..a17ce9f 100644 --- a/scribo/src/binarization/ppm_sauvola_ms_fg.cc +++ b/scribo/src/binarization/ppm_sauvola_ms_fg.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -31,7 +31,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/preprocessing/split_bg_fg.hh> @@ -122,7 +122,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/ppm_sauvola_ms_split.cc b/scribo/src/binarization/ppm_sauvola_ms_split.cc index bb98b38..92a0817 100644 --- a/scribo/src/binarization/ppm_sauvola_ms_split.cc +++ b/scribo/src/binarization/ppm_sauvola_ms_split.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -29,7 +29,6 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> #include <scribo/binarization/sauvola_ms_split.hh> #include <scribo/debug/usage.hh> diff --git a/scribo/src/binarization/sauvola.cc b/scribo/src/binarization/sauvola.cc index 0273071..e8047ab 100644 --- a/scribo/src/binarization/sauvola.cc +++ b/scribo/src/binarization/sauvola.cc @@ -29,7 +29,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola.hh> #include <scribo/debug/usage.hh> @@ -77,7 +77,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input, mln::fun::v2v::rgb_to_luma<value::int_u8>()); image2d<bool> out = scribo::binarization::sauvola(input_1_gl, w, k); diff --git a/scribo/src/binarization/sauvola_debug.cc b/scribo/src/binarization/sauvola_debug.cc index f723851..0c335a8 100644 --- a/scribo/src/binarization/sauvola_debug.cc +++ b/scribo/src/binarization/sauvola_debug.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -32,7 +32,7 @@ #include <mln/data/convert.hh> #include <mln/data/saturate.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/local_threshold.hh> #include <scribo/binarization/sauvola.hh> @@ -116,7 +116,7 @@ int main(int argc, char *argv[]) image2d<value::int_u8> gima = data::transform(input, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); image2d<bool> diff --git a/scribo/src/binarization/sauvola_ms.cc b/scribo/src/binarization/sauvola_ms.cc index 541b9e5..6d60ab4 100644 --- a/scribo/src/binarization/sauvola_ms.cc +++ b/scribo/src/binarization/sauvola_ms.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -31,7 +31,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -115,7 +115,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/sauvola_ms_debug.cc b/scribo/src/binarization/sauvola_ms_debug.cc index 6bf9837..70c1a9a 100644 --- a/scribo/src/binarization/sauvola_ms_debug.cc +++ b/scribo/src/binarization/sauvola_ms_debug.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -29,7 +29,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -124,7 +124,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize. diff --git a/scribo/src/binarization/sauvola_ms_fg.cc b/scribo/src/binarization/sauvola_ms_fg.cc index 4227db4..7ff9321 100644 --- a/scribo/src/binarization/sauvola_ms_fg.cc +++ b/scribo/src/binarization/sauvola_ms_fg.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -31,7 +31,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/preprocessing/split_bg_fg.hh> @@ -102,7 +102,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/text_in_picture.cc b/scribo/src/text_in_picture.cc index e2f30de..77cf7b6 100644 --- a/scribo/src/text_in_picture.cc +++ b/scribo/src/text_in_picture.cc @@ -40,7 +40,7 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/subsampling/antialiased.hh> @@ -236,7 +236,8 @@ int main(int argc, char* argv[]) std::cout << "** Using split_bg_fg" << std::endl; image2d<value::rgb8> fg = preprocessing::split_bg_fg(input_rgb, lambda, 32).second(); - intensity_ima = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + intensity_ima = data::transform(fg, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); t_ = timer_; std::cout << "Foreground extracted. " << t_ << std::endl; @@ -253,7 +254,7 @@ int main(int argc, char* argv[]) timer_.start(); std::cout << "** Using data::transform(intensity)" << std::endl; intensity_ima = data::transform(input_rgb, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); t_ = timer_; std::cout << "Intensity image " << t_ << std::endl; } diff --git a/scribo/src/text_in_picture_neg.cc b/scribo/src/text_in_picture_neg.cc index c1a4317..8d818e0 100644 --- a/scribo/src/text_in_picture_neg.cc +++ b/scribo/src/text_in_picture_neg.cc @@ -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. // @@ -48,8 +49,6 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> - #include <mln/data/wrap.hh> #include <mln/draw/box.hh> diff --git a/scribo/src/text_recognition_in_picture.cc b/scribo/src/text_recognition_in_picture.cc index a5f93a7..8b76f48 100644 --- a/scribo/src/text_recognition_in_picture.cc +++ b/scribo/src/text_recognition_in_picture.cc @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -51,7 +51,7 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/data/wrap.hh> @@ -253,7 +253,8 @@ int main(int argc, char* argv[]) // Extract foreground image2d<value::rgb8> fg = preprocessing::split_bg_fg(input_rgb, lambda, 32).second(); - intensity_ima = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + intensity_ima = data::transform(fg, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // // Perform an initial rotation if needed. // // input_rgb = geom::rotate(input_rgb, -45, literal::black); -- 1.5.6.5

1 0

last-svn-commit-780-gb37a853 mln/labeling/fill_holes.hh: Improve speed.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

--- milena/ChangeLog | 4 ++++ milena/mln/labeling/fill_holes.hh | 17 +++++++++-------- 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/milena/ChangeLog b/milena/ChangeLog index edefc8e..2eed916 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * mln/labeling/fill_holes.hh: Improve speed. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * mln/fun/v2v/rgb_to_luma.hh: New function for grayscale conversion. diff --git a/milena/mln/labeling/fill_holes.hh b/milena/mln/labeling/fill_holes.hh index e76c489..5e487d9 100644 --- a/milena/mln/labeling/fill_holes.hh +++ b/milena/mln/labeling/fill_holes.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2007, 2008, 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2007, 2008, 2009, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -33,6 +34,8 @@ # include <mln/labeling/background.hh> # include <mln/labeling/compute.hh> +# include <mln/data/transform.hh> + # include <mln/core/image/dmorph/image_if.hh> # include <mln/accu/math/count.hh> @@ -57,7 +60,7 @@ namespace mln /// \see mln::labeling::background /// template <typename I, typename N, typename L> - I + mln_concrete(I) fill_holes(const Image& input, const Neighborhood<N>& nbh, L& nlabels); @@ -66,7 +69,7 @@ namespace mln template <typename I, typename N, typename L> inline - I + mln_concrete(I) fill_holes(const Image& input, const Neighborhood<N>& nbh, L& nlabels) { @@ -77,10 +80,6 @@ namespace mln mln_precondition(exact(input).is_valid()); mln_precondition(exact(nbh).is_valid()); - mln_ch_value(I, bool) output; - initialize(output, input); - data::fill(output, false); - mln_ch_value(I, L) lbls = labeling::background(input, nbh, nlabels); accu::math::count<mln_value(I)> a_; @@ -99,7 +98,9 @@ namespace mln } } - data::fill((output | (pw::value(lbls) != bg_lbl)).rw(), true); + util::array<bool> bg_relbl(arr.nelements(), true); + bg_relbl(bg_lbl) = false; + mln_ch_value(I, bool) output = data::transform(lbls, bg_relbl); trace::exiting("labeling::fill_holes"); return output; -- 1.5.6.5

1 0

last-svn-commit-779-g5127a7b mln/fun/v2v/rgb_to_luma.hh: New function for grayscale conversion.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

--- milena/ChangeLog | 5 +++ .../achromatism.hh => mln/fun/v2v/rgb_to_luma.hh} | 32 ++++++++++++------- 2 files changed, 25 insertions(+), 12 deletions(-) copy milena/{sandbox/green/mln/fun/v2v/achromatism.hh => mln/fun/v2v/rgb_to_luma.hh} (67%) diff --git a/milena/ChangeLog b/milena/ChangeLog index c4bed2c..edefc8e 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,3 +1,8 @@ +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + + * mln/fun/v2v/rgb_to_luma.hh: New function for grayscale + conversion. + 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> * mln/util/array.hh: Add last() method. diff --git a/milena/sandbox/green/mln/fun/v2v/achromatism.hh b/milena/mln/fun/v2v/rgb_to_luma.hh similarity index 67% copy from milena/sandbox/green/mln/fun/v2v/achromatism.hh copy to milena/mln/fun/v2v/rgb_to_luma.hh index 72b545c..304a798 100644 --- a/milena/sandbox/green/mln/fun/v2v/achromatism.hh +++ b/milena/mln/fun/v2v/rgb_to_luma.hh @@ -1,4 +1,4 @@ -// Copyright (C) 2008, 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -23,10 +23,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -#ifndef MLN_FUN_V2V_ACHROMATISM_HH -# define MLN_FUN_V2V_ACHROMATISM_HH +#ifndef MLN_FUN_V2V_RGB_TO_LUMA_HH +# define MLN_FUN_V2V_RGB_TO_LUMA_HH -# include <mln/value/rgb8.hh> +# include <mln/value/rgb.hh> namespace mln { @@ -37,22 +37,30 @@ namespace mln namespace v2v { - struct achromatism : public Function_v2v< achromatism > + template <typename T_luma> + struct rgb_to_luma : public Function_v2v< rgb_to_luma<T_luma> > { - typedef float result; + typedef T_luma result; + + template <typename T_rgb> + T_luma operator()(const T_rgb& rgb) const; - float operator()(const value::rgb8 rgb) const; }; + # ifndef MLN_INCLUDE_ONLY - float achromatism::operator()(const value::rgb8 rgb) const + template <typename T_luma> + template <typename T_rgb> + inline + T_luma + rgb_to_luma<T_luma>::operator()(const T_rgb& rgb) const { - return (math::abs(rgb.red() - rgb.green()) - + math::abs(rgb.red() - rgb.blue()) - + math::abs(rgb.green() - rgb.blue()))/3.0; + float luma = 0.299 * rgb.red() + 0.587 * rgb.green() + 0.114 * rgb.blue(); + return unsigned(luma + 0.49999); } + # endif // !MLN_INCLUDE_ONLY } // end of namespace fun::v2v @@ -61,4 +69,4 @@ namespace mln } // end of namespace mln -#endif // ! MLN_FUN_V2V_ACHROMATISM_HH +#endif // ! MLN_FUN_V2V_RGB_TO_LUMA_HH -- 1.5.6.5

1 0

last-svn-commit-778-gb1fe7b8 scribo/io/xml/load.hh: New XML loader.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

--- scribo/ChangeLog | 4 + scribo/scribo/io/xml/load.hh | 525 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 529 insertions(+), 0 deletions(-) create mode 100644 scribo/scribo/io/xml/load.hh diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 6571137..cf51ca8 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/io/xml/load.hh: New XML loader. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make XML output more flexible. * scribo/core/component_info.hh, diff --git a/scribo/scribo/io/xml/load.hh b/scribo/scribo/io/xml/load.hh new file mode 100644 index 0000000..e0f4548 --- /dev/null +++ b/scribo/scribo/io/xml/load.hh @@ -0,0 +1,525 @@ +// 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/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_LOAD_HH +# define SCRIBO_IO_XML_LOAD_HH + +/// \file +/// +/// \brief Load document information from XML. + +# include <QtXml> + +# include <libgen.h> +# include <fstream> +# include <sstream> + +# include <map> + +# include <mln/core/image/image2d.hh> + +# include <mln/data/wrap.hh> +# include <mln/value/int_u8.hh> +# include <mln/io/pgm/save.hh> +# include <mln/io/pbm/save.hh> + +# include <scribo/core/document.hh> +# include <scribo/core/component_set.hh> +# include <scribo/core/line_set.hh> +# include <scribo/core/line_info.hh> + +# include <scribo/convert/from_base64.hh> + +namespace scribo +{ + + namespace io + { + + namespace xml + { + using namespace mln; + + /*! \brief Load document information from XML. + + We use a XML Schema part of the PAGE (Page Analysis and Ground + truth Elements) image representation framework. + + This schema was used in the Page Segmentation COMPetition + (PSCOMP) for ICDAR 2009. + + Its XSD file is located here: + http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16/pagecontent… + + */ + template <typename L> + void + load(document<L>& doc, const std::string& input_name); + + +# ifndef MLN_INCLUDE_ONLY + + namespace internal + { + + + enum Mode + { + None, + ComponentSet, + ComponentInfo, + LabeledImage, + SeparatorsImage, + ObjectLinks, + ObjectGroups, + Point, + Link, + Group, + Line, + LineLinks, + LineLink, + TextData, + TextRegion, + CompIdList, + CompId, + Page + }; + + + struct ModeData + { + const char *name; + Mode mode; + }; + + + static const ModeData mode_data[] = { + { "component_set", ComponentSet }, + { "component_info", ComponentInfo }, + { "labeled_image", LabeledImage }, + { "separators_image", SeparatorsImage }, + { "object_links", ObjectLinks }, + { "object_groups", ObjectGroups }, + { "point", Point }, + { "link", Link }, + { "group", Group }, + { "line", Line }, + { "line_links", LineLinks }, + { "line_link", LineLink }, + { "text_data", TextData }, + { "text_region", TextRegion }, + { "compid_list", CompIdList }, + { "compid", CompId }, + { "page", Page }, + { 0, None } + }; + + + template <typename L> + class xml_handler : public QXmlDefaultHandler + { + + typedef mln_ch_value(L,bool) B; + + public: + xml_handler() : current_paragraph_id(1) { lines_data.append(line_info<L>()); } // line info id starts from 1. + + virtual + bool + startElement(const QString& /*namespaceURI*/, const QString& /*localName*/, + const QString& qName, const QXmlAttributes& atts ) + { + mode.push(find_mode(qName)); + + switch (mode.top()) + { + + // Component Set + case ComponentSet: + { + comp_set_data = new scribo::internal::component_set_data<L>(); + // qDebug() << qName << " - atts.value(\"nelements\").toInt() = " << atts.value("nelements").toInt();; + comp_set_data->soft_init(atts.value("nelements").toInt()); + } + break; + + + // Component Info + case ComponentInfo: + { + component_info info(atts.value("id").toInt(), + mln::make::box2d(atts.value("pmin_y").toInt(), + atts.value("pmin_x").toInt(), + atts.value("pmax_y").toInt(), + atts.value("pmax_x").toInt()), + mln::point2d(atts.value("mass_center_y").toInt(), + atts.value("mass_center_x").toInt()), + atts.value("card").toInt()); + + + info.update_tag(component::str2tag(atts.value("tag").toUtf8().constData())); + info.update_type(component::str2type(atts.value("type").toUtf8().constData())); + + comp_set_data->infos_.append(info); + } + break; + + + // Object links + case ObjectLinks: + { + // qDebug() << "object_links created"; + links = object_links<L>(components); + } + break; + + + // Object groups + case ObjectGroups: + { + // qDebug() << "object_groups created"; + groups = object_groups<L>(links); + } + break; + + + // Text data + case TextData: + { + // qDebug() << "TextData"; + + // Reserve space for line data. + lines_data.resize(atts.value("nlines").toInt() + 1); + // qDebug() << "line_set created"; + lines = line_set<L>(groups, lines_data); + llinks = line_links<L>(lines); + par_data = new scribo::internal::paragraph_set_data<L>(llinks, atts.value("nparagraphs").toInt()); + // std::cout << par_data->pars_.nelements() << " - " << llinks.nelements() << " - " << lines.nelements() << std::endl; + } + break; + + + // Text Region + case TextRegion: + { + // qDebug() << "TextRegion"; + + current_paragraph = paragraph_info<L>(llinks); + } + break; + + + // Line link + case LineLink: + { + llinks(atts.value("from").toInt()) = atts.value("to").toInt(); + } + break; + + + // Line + case Line: + { + current_line_id = atts.value("id").toInt(); + + line_data = new scribo::internal::line_info_data<L>(lines, mln::util::array<component_id_t>()); + + line_data->holder_ = lines; + line_data->text_ = atts.value("text").toUtf8().constData(); + + line_data->hidden_ = false; + line_data->tag_ = static_cast<line::Tag>(atts.value("tag").toInt()); + + line_data->baseline_ = atts.value("baseline").toInt(); + line_data->meanline_ = atts.value("meanline").toInt(); + line_data->x_height_ = atts.value("x_height").toInt(); + line_data->d_height_ = atts.value("d_height").toInt(); + line_data->a_height_ = atts.value("a_height").toInt(); + line_data->char_space_ = atts.value("kerning").toInt(); + line_data->char_width_ = atts.value("char_width").toInt(); + line_data->word_space_ = 0; + + line_data->reading_direction_ = line::LeftToRight; + line_data->type_ = line::str2type(atts.value("txt_text_type").toAscii().constData()); + line_data->reverse_video_ = (atts.value("txt_reverse_video") == "false" ? false : true); + line_data->orientation_ = 0; + line_data->reading_orientation_ = atts.value("txt_reading_orientation").toInt(); + line_data->indented_ = (atts.value("txt_indented") == "false" ? false : true); + + bbox.init(); + } + break; + + + // CompIdList + case CompIdList: + { + + } + break; + + + // CompId + case CompId: + { + line_data->components_.append(atts.value("value").toInt()); + } + break; + + + // Point + case Point: + { + point2d p(atts.value("y").toInt(), atts.value("x").toInt()); + bbox.take(p); + } + break; + + + // Labeled Image + case LabeledImage: + { + width = atts.value("width").toInt(); + height = atts.value("height").toInt(); + comp_set_data->ima_ = L(mln::make::box2d(height, width), 0); // No border + } + break; + + + // Separator Image + case SeparatorsImage: + { + width = atts.value("width").toInt(); + height = atts.value("height").toInt(); + comp_set_data->separators_ = B(mln::make::box2d(height, width), 0); // No border + } + break; + + + // Link + case Link: + { + links(atts.value("from").toInt()) = atts.value("to").toInt(); + } + break; + + + // Group + case Group: + { + groups(atts.value("object_id").toInt()) = atts.value("group_id").toInt(); + } + break; + + + // DEFAULT + default: + ; + } + + return true; + } + + + virtual + bool + endElement(const QString& /*namespaceURI*/, const QString& /*localName*/, const QString& /*qName*/) + { + switch(mode.top()) + { + // Component set + case ComponentSet: + { + // qDebug() << "Component set done"; + components = component_set<L>(comp_set_data); + } + break; + + // Line + case Line: + { + // qDebug() << "Line done"; + line_data->bbox_ = bbox; + lines_data(current_line_id) = line_info<L>(current_line_id, line_data); + lines_data(current_line_id).update_ebbox(); + + // Add this line to the current paragraph. + current_paragraph.add_line(lines_data(current_line_id)); + } + break; + + // TextRegion + case TextRegion: + { + // qDebug() << TextRegion; + par_data->pars_(current_paragraph_id++) = current_paragraph; + } + break; + + // Page + case Page: + { + // qDebug() << "Page done"; + lines.update_line_data_(lines_data); + parset = paragraph_set<L>(par_data); + } + break; + + // DEFAULT + default: + ; + + } + + mode.pop(); + return true; + } + + + + bool characters(const QString & ch) + { + switch (mode.top()) + { + case LabeledImage: + { + scribo::convert::from_base64(ch, comp_set_data->ima_); + } + break; + + case SeparatorsImage: + { + scribo::convert::from_base64(ch, comp_set_data->separators_); + } + break; + + default: + ; + } + + return true; + } + + +// private: // Methods + + Mode find_mode(const QString& qName) + { + for (int i = 0; mode_data[i].name; ++i) + if (mode_data[i].name == qName) + return mode_data[i].mode; + return None; + } + + +// private: // Attributes + + QStack<Mode> mode; + + // Shape + accu::shape::bbox<point2d> bbox; + + unsigned width; + unsigned height; + + // Components + mln::util::tracked_ptr<scribo::internal::component_set_data<L> > comp_set_data; + component_set<L> components; + + object_links<L> links; + object_groups<L> groups; + + // Lines + unsigned current_line_id; + scribo::internal::line_info_data<L> *line_data; + + line_links<L> llinks; + + unsigned current_paragraph_id; + paragraph_info<L> current_paragraph; + scribo::internal::paragraph_set_data<L> *par_data; + paragraph_set<L> parset; + + mln::util::array<line_info<L> > lines_data; + line_set<L> lines; + }; + + + + + + + + + + + template <typename L> + void + load_extended(document<L>& doc, + const std::string& output_name) + { + xml_handler<L> handler; + QXmlSimpleReader reader; + reader.setContentHandler(&handler); + + QFile file(output_name.c_str()); + if (!file.open(QFile::ReadOnly | QFile::Text)) + { + qDebug() << "Cannot read file"; + return; + } + + QXmlInputSource xmlInputSource(&file); + if (reader.parse(xmlInputSource)) + qDebug() << "Loaded successfuly"; + + doc.set_paragraphs(handler.parset); + } + + } // end of namespace scribo::io::xml::internal + + + // FACADE + + template <typename L> + void + load(document<L>& doc, + const std::string& output_name) + { + internal::load_extended(doc, output_name); + } + + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + + +#endif // ! SCRIBO_IO_XML_LOAD_HH + -- 1.5.6.5

1 0

last-svn-commit-776-g37420bc Set component type during component extraction.
by Guillaume Lazzara 01 Mar '11

01 Mar '11

* scribo/core/component_info.hh, * scribo/core/component_set.hh, * scribo/core/document.hh, * scribo/core/tag/component.hh, * scribo/primitive/extract/components.hh, * scribo/primitive/identify.hh: Explicitly set component type to Separator when extracting separator components. --- scribo/ChangeLog | 12 ++++++ scribo/scribo/core/component_info.hh | 8 ++- scribo/scribo/core/component_set.hh | 50 ++++++++++++++++--------- scribo/scribo/core/document.hh | 6 ++- scribo/scribo/core/tag/component.hh | 20 ++++++--- scribo/scribo/primitive/extract/components.hh | 18 ++++++--- scribo/scribo/primitive/identify.hh | 2 +- 7 files changed, 79 insertions(+), 37 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 8b7ad7f..330338a 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,15 @@ +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + + Set component type during component extraction. + + * scribo/core/component_info.hh, + * scribo/core/component_set.hh, + * scribo/core/document.hh, + * scribo/core/tag/component.hh, + * scribo/primitive/extract/components.hh, + * scribo/primitive/identify.hh: Explicitly set component type to + Separator when extracting separator components. + 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> Add new tools in Scribo. diff --git a/scribo/scribo/core/component_info.hh b/scribo/scribo/core/component_info.hh index 1b03318..6fc73f8 100644 --- a/scribo/scribo/core/component_info.hh +++ b/scribo/scribo/core/component_info.hh @@ -53,7 +53,8 @@ namespace scribo component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, - unsigned card); + unsigned card, + component::Type type = component::Undefined); component_id_t id() const; const mln::box2d& bbox() const; @@ -101,9 +102,10 @@ namespace scribo component_info::component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, - unsigned card) + unsigned card, + component::Type type) : id_(id), bbox_(bbox), mass_center_(mass_center), card_(card), - tag_(component::None), type_(component::Undefined) + tag_(component::None), type_(type) { } diff --git a/scribo/scribo/core/component_set.hh b/scribo/scribo/core/component_set.hh index 7ddcf16..442e8d6 100644 --- a/scribo/scribo/core/component_set.hh +++ b/scribo/scribo/core/component_set.hh @@ -86,16 +86,20 @@ namespace scribo component_set_data(); component_set_data(const L& ima, const mln_value(L)& ncomps); component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs); + const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs); + const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); component_set_data(const L& ima, const mln_value(L)& ncomps, const mln::util::array<scribo::component_info>& infos); - void fill_infos(const mln::util::array<pair_accu_t>& attribs); + void fill_infos(const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); - void fill_infos(const mln::util::array<pair_data_t>& attribs); + void fill_infos(const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); // Useful while constructing incrementaly (XML loading). void soft_init(const mln_value(L) ncomps); @@ -141,10 +145,12 @@ namespace scribo /// Constructor from an image \p ima, the number of labels \p ncomps and /// attributes values (bounding box and mass center). component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs); + const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs); + const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); /// @} /// Return the component count. @@ -284,26 +290,28 @@ namespace scribo inline component_set_data<L>::component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs) + const mln::util::array<pair_accu_t>& attribs, + component::Type type) : ima_(ima), ncomps_(ncomps) { initialize(separators_, ima); // FIXME: do we really want that? mln::data::fill(separators_, false); - fill_infos(attribs); + fill_infos(attribs, type); } template <typename L> inline component_set_data<L>::component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs) + const mln::util::array<pair_data_t>& attribs, + component::Type type) : ima_(ima), ncomps_(ncomps) { initialize(separators_, ima); // FIXME: do we really want that? mln::data::fill(separators_, false); - fill_infos(attribs); + fill_infos(attribs, type); } template <typename L> @@ -321,7 +329,8 @@ namespace scribo template <typename L> inline void - component_set_data<L>::fill_infos(const mln::util::array<pair_accu_t>& attribs) + component_set_data<L>::fill_infos(const mln::util::array<pair_accu_t>& attribs, + component::Type type) { typedef mln_site(L) P; @@ -331,7 +340,8 @@ namespace scribo for_all_comp_data(i, attribs) { component_info info(i, attribs[i].first(), - attribs[i].second(), attribs[i].second_accu().nsites()); + attribs[i].second(), attribs[i].second_accu().nsites(), + type); infos_.append(info); } } @@ -339,7 +349,8 @@ namespace scribo template <typename L> inline void - component_set_data<L>::fill_infos(const mln::util::array<pair_data_t>& attribs) + component_set_data<L>::fill_infos(const mln::util::array<pair_data_t>& attribs, + component::Type type) { typedef mln_site(L) P; @@ -349,7 +360,8 @@ namespace scribo for_all_comp_data(i, attribs) { component_info info(i, attribs[i].first, - attribs[i].second.first, attribs[i].second.second); + attribs[i].second.first, attribs[i].second.second, + type); infos_.append(info); } } @@ -397,9 +409,10 @@ namespace scribo template <typename L> inline component_set<L>::component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs) + const mln::util::array<pair_accu_t>& attribs, + component::Type type) { - data_ = new internal::component_set_data<L>(ima, ncomps, attribs); + data_ = new internal::component_set_data<L>(ima, ncomps, attribs, type); } @@ -407,9 +420,10 @@ namespace scribo inline component_set<L>::component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs) + const mln::util::array<pair_data_t>& attribs, + component::Type type) { - data_ = new internal::component_set_data<L>(ima, ncomps, attribs); + data_ = new internal::component_set_data<L>(ima, ncomps, attribs, type); } diff --git a/scribo/scribo/core/document.hh b/scribo/scribo/core/document.hh index e5ac825..ef0869e 100644 --- a/scribo/scribo/core/document.hh +++ b/scribo/scribo/core/document.hh @@ -297,7 +297,8 @@ namespace scribo mln_value(L) ncomps; whitespace_seps_comps_ = primitive::extract::components(whitespace_seps, - mln::c8(), ncomps); + mln::c8(), ncomps, + component::WhitespaceSeparator); } @@ -333,7 +334,8 @@ namespace scribo mln_value(L) ncomps; line_seps_comps_ = primitive::extract::components(line_seps, - mln::c8(), ncomps); + mln::c8(), ncomps, + component::LineSeparator); } diff --git a/scribo/scribo/core/tag/component.hh b/scribo/scribo/core/tag/component.hh index 10b86a6..7cd2ede 100644 --- a/scribo/scribo/core/tag/component.hh +++ b/scribo/scribo/core/tag/component.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -55,7 +55,8 @@ namespace scribo { Undefined = 0, Character, - Separator, + LineSeparator, + WhitespaceSeparator, Noise, Punctuation, Image @@ -116,8 +117,11 @@ namespace scribo case Character: str = "Character"; break; - case Separator: - str = "Separator"; + case LineSeparator: + str = "LineSeparator"; + break; + case WhitespaceSeparator: + str = "WhitespaceSeparator"; break; case Noise: str = "Noise"; @@ -139,8 +143,10 @@ namespace scribo { if (str == "Character") return Character; - else if (str == "Separator") - return Separator; + else if (str == "LineSeparator") + return LineSeparator; + else if (str == "WhitespaceSeparator") + return WhitespaceSeparator; else if (str == "Noise") return Noise; else if (str == "Punctuation") diff --git a/scribo/scribo/primitive/extract/components.hh b/scribo/scribo/primitive/extract/components.hh index 4994d4b..849dd7b 100644 --- a/scribo/scribo/primitive/extract/components.hh +++ b/scribo/scribo/primitive/extract/components.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2009, 2011 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of Olena. // @@ -68,6 +69,7 @@ namespace scribo /// and background to 'false'. /// \param[in] nbh A neighborhood to be used for labeling. /// \param[in,out] ncomponents Will store the numbers of components found. + /// \param[in] type The default component type set to components. /// /// \return An image of labeled components. // @@ -75,7 +77,8 @@ namespace scribo inline component_set<mln_ch_value(I,V)> components(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents); + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type = component::Undefined); # ifndef MLN_INCLUDE_ONLY @@ -88,7 +91,8 @@ namespace scribo inline void components_tests(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents) + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type) { mlc_equal(mln_value(I),bool)::check(); // mlc_is_a(V, mln::value::Symbolic)::check(); @@ -97,6 +101,7 @@ namespace scribo (void) input; (void) nbh; (void) ncomponents; + (void) type; } @@ -107,11 +112,12 @@ namespace scribo inline component_set<mln_ch_value(I,V)> components(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents) + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type = component::Undefined) { trace::entering("scribo::components"); - internal::components_tests(input, nbh, ncomponents); + internal::components_tests(input, nbh, ncomponents, type); typedef mln_ch_value(I,V) L; typedef mln::accu::shape::bbox<mln_site(L)> bbox_accu_t; @@ -129,7 +135,7 @@ namespace scribo pair_accu_t()); component_set<L> - output(results.first(), ncomponents, results.second().second()); + output(results.first(), ncomponents, results.second().second(), type); trace::exiting("scribo::components"); return output; diff --git a/scribo/scribo/primitive/identify.hh b/scribo/scribo/primitive/identify.hh index 81a7d16..1bed712 100644 --- a/scribo/scribo/primitive/identify.hh +++ b/scribo/scribo/primitive/identify.hh @@ -61,7 +61,7 @@ namespace scribo std::swap(min, max); if (max/min > 10) - output(c).update_type(component::Separator); + output(c).update_type(component::LineSeparator); } mln::trace::exiting("scribo::primitive::identify"); -- 1.5.6.5

1 0