* scribo/debug/logger.hh: New. * scribo/toolchain/internal/content_in_doc_functor.hh, * scribo/toolchain/internal/text_in_doc_functor.hh, * scribo/toolchain/nepomuk/text_extraction.hh, * scribo/toolchain/text_in_doc.hh, * src/content_in_doc.cc, * src/pbm_text_in_doc.cc, * src/primitive/extract/separators_nonvisible.cc Make use of that class for saving debug images. --- scribo/ChangeLog | 15 ++ scribo/scribo/toolchain/content_in_doc.hh | 10 +- .../toolchain/internal/content_in_doc_functor.hh | 202 ++++++++++++------- .../toolchain/internal/text_in_doc_functor.hh | 107 ++++++----- scribo/scribo/toolchain/nepomuk/text_extraction.hh | 9 +- scribo/scribo/toolchain/text_in_doc.hh | 11 +- scribo/src/content_in_doc.cc | 21 +-- scribo/src/pbm_text_in_doc.cc | 16 +- .../src/primitive/extract/separators_nonvisible.cc | 8 +- 9 files changed, 234 insertions(+), 165 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index a377ba0..1e9b57e 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -4,6 +4,21 @@ * scribo/debug/logger.hh: New. + * scribo/toolchain/internal/content_in_doc_functor.hh, + * scribo/toolchain/internal/text_in_doc_functor.hh, + * scribo/toolchain/nepomuk/text_extraction.hh, + * scribo/toolchain/text_in_doc.hh, + * src/content_in_doc.cc, + * src/pbm_text_in_doc.cc, + * src/primitive/extract/separators_nonvisible.cc Make use of that + class for saving debug images. + +2011-03-14 Guillaume Lazzara <z@lrde.epita.fr> + + Introduce a logger class. + + * scribo/debug/logger.hh: New. + 2011-03-14 Guillaume Lazzara <z@lrde.epita.fr> Introduce new anchors. diff --git a/scribo/scribo/toolchain/content_in_doc.hh b/scribo/scribo/toolchain/content_in_doc.hh index f2938d9..4469afa 100644 --- a/scribo/scribo/toolchain/content_in_doc.hh +++ b/scribo/scribo/toolchain/content_in_doc.hh @@ -45,10 +45,10 @@ namespace scribo document<mln_ch_value(I, def::lbl_type)> content_in_doc(const Image<I>& input, const Image<J>& input_preproc, bool denoise, - const std::string& language = std::string("eng"), bool find_line_seps = true, bool find_whitespace_seps = true, - bool debug = false); + bool enable_ocr = true, + const std::string& language = std::string("eng")); # ifndef MLN_INCLUDE_ONLY @@ -58,10 +58,10 @@ namespace scribo document<mln_ch_value(I, def::lbl_type)> content_in_doc(const Image<I>& input, const Image<J>& input_preproc, bool denoise, - const std::string& language = std::string("eng"), bool find_line_seps = true, bool find_whitespace_seps = true, - bool debug = false) + bool enable_ocr = true, + const std::string& language = std::string("eng")) { mln_precondition(input.is_valid()); mln_precondition(input_preproc.is_valid()); @@ -70,8 +70,8 @@ namespace scribo f.enable_denoising = denoise; f.enable_line_seps = find_line_seps; f.enable_whitespace_seps = find_whitespace_seps; - f.enable_debug = debug; f.ocr_language = language; + f.enable_ocr = enable_ocr; document<mln_ch_value(I, def::lbl_type)> doc = f(input, input_preproc); diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index 3e6668f..624a54e 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -27,8 +27,6 @@ #ifndef SCRIBO_TOOLCHAIN_INTERNAL_CONTENT_IN_DOC_FUNCTOR_HH # define SCRIBO_TOOLCHAIN_INTERNAL_CONTENT_IN_DOC_FUNCTOR_HH -# include <mln/io/ppm/save.hh> - # include <scribo/core/def/lbl_type.hh> # include <scribo/core/document.hh> # include <scribo/core/line_set.hh> @@ -37,6 +35,8 @@ # include <scribo/primitive/extract/non_text.hh> # include <scribo/primitive/extract/components.hh> # include <scribo/primitive/extract/separators.hh> +# include <scribo/primitive/extract/vertical_separators.hh> +# include <scribo/primitive/extract/horizontal_separators.hh> # include <scribo/primitive/extract/separators_nonvisible.hh> # include <scribo/primitive/identify.hh> @@ -62,8 +62,9 @@ # include <scribo/make/debug_filename.hh> -# include <scribo/debug/save_bboxes_image.hh> -# include <scribo/debug/save_linked_bboxes_image.hh> +# include <scribo/debug/decision_image.hh> +# include <scribo/debug/bboxes_image.hh> +# include <scribo/debug/linked_bboxes_image.hh> # include <scribo/debug/bboxes_enlarged_image.hh> # include <scribo/debug/mean_and_base_lines_image.hh> # include <scribo/debug/looks_like_a_text_line_image.hh> @@ -87,7 +88,7 @@ namespace scribo struct content_in_doc_functor : public Toolchain_Functor { - typedef value::label<30> V; + typedef scribo::def::lbl_type V; typedef mln_ch_value(I,V) L; content_in_doc_functor(const char *doc_filename); @@ -112,7 +113,6 @@ namespace scribo bool enable_line_seps; bool enable_whitespace_seps; bool enable_ocr; - bool enable_debug; bool save_doc_as_xml; scribo::io::xml::Format xml_format; @@ -138,7 +138,6 @@ namespace scribo enable_line_seps(true), enable_whitespace_seps(true), enable_ocr(true), - enable_debug(false), save_doc_as_xml(false), xml_format(scribo::io::xml::PageExtended), ocr_language("eng"), @@ -209,21 +208,30 @@ namespace scribo on_progress(); } - if (enable_debug) + + // Debug +# ifndef SCRIBO_NDEBUG + if (enable_whitespace_seps) + debug::logger().log_image(debug::AuxiliaryResults, + whitespaces, "whitespaces"); + + // Debug + if (enable_line_seps) { - if (enable_whitespace_seps) - mln::io::pbm::save(whitespaces, - scribo::make::debug_filename("whitespaces.pbm")); + debug::logger().log_image(debug::AuxiliaryResults, + doc.vline_seps(), + "vseparators"); - if (enable_line_seps) - { - mln::io::pbm::save(separators, - scribo::make::debug_filename("vseparators.pbm")); + debug::logger().log_image(debug::AuxiliaryResults, + doc.hline_seps(), + "hseparators"); - mln::io::pbm::save(input_cleaned, - scribo::make::debug_filename("input_wo_vseparators.pbm")); - } + debug::logger().log_image(debug::AuxiliaryResults, + input_cleaned, + "input_wo_separators"); } +# endif // ! SCRIBO_NDEBUG + // Denoise if (enable_denoising) @@ -232,9 +240,11 @@ namespace scribo input_cleaned = preprocessing::denoise_fg(input_cleaned, c8(), 3); - if (enable_debug) - mln::io::pbm::save(input_cleaned, - scribo::make::debug_filename("denoised.pbm")); + // Debug +# ifndef SCRIBO_NDEBUG + debug::logger().log_image(debug::AuxiliaryResults, + input_cleaned, "denoised"); +# endif // ! SCRIBO_NDEBUG on_progress(); } @@ -244,7 +254,8 @@ namespace scribo V ncomponents; component_set<L> - components = scribo::primitive::extract::components(input_cleaned, c8(), + components = scribo::primitive::extract::components(input_cleaned, + c8(), ncomponents); on_progress(); @@ -258,9 +269,12 @@ namespace scribo doc.set_whitespace_separators(whitespaces); } - if (enable_debug) - mln::io::pbm::save(components.separators(), - scribo::make::debug_filename("all_separators.pbm")); + // Debug +# ifndef SCRIBO_NDEBUG + debug::logger().log_image(debug::AuxiliaryResults, + components.separators(), + "all_separators"); +# endif // ! SCRIBO_NDEBUG on_new_progress_label("Filtering components"); @@ -274,24 +288,34 @@ namespace scribo on_new_progress_label("Linking objects..."); object_links<L> left_link - = primitive::link::with_single_left_link_dmax_ratio(components, - primitive::link::internal::dmax_width_and_height(1), - anchor::MassCenter); - object_links<L> right_link - = primitive::link::with_single_right_link_dmax_ratio(components, - primitive::link::internal::dmax_width_and_height(1), - anchor::MassCenter); + = primitive::link::with_single_left_link_dmax_ratio( + components, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); - if (enable_debug) + object_links<L> right_link + = primitive::link::with_single_right_link_dmax_ratio( + components, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); + + // Debug +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { - debug::save_linked_bboxes_image(processed_image, left_link, right_link, - literal::blue, - literal::cyan, - literal::yellow, - literal::green, - anchor::MassCenter, - scribo::make::debug_filename("object_links.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + debug::linked_bboxes_image(processed_image, + left_link, + right_link, + literal::blue, + literal::cyan, + literal::yellow, + literal::green, + anchor::MassCenter), + "object_links"); } +# endif // ! SCRIBO_NDEBUG // Validating left and right links. @@ -309,16 +333,20 @@ namespace scribo = filter::object_links_bbox_h_ratio(merged_links, 2.5f); - if (enable_debug) +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { mln_ch_value(I,value::rgb8) hratio_decision_image = scribo::debug::decision_image(processed_image, merged_links, hratio_filtered_links, anchor::MassCenter); - mln::io::ppm::save(hratio_decision_image, - scribo::make::debug_filename("hratio_links_decision_image.ppm")); + // Debug + debug::logger().log_image(debug::AuxiliaryResults, + hratio_decision_image, + "hratio_links_decision_image"); } +# endif // ! SCRIBO_NDEBUG on_progress(); @@ -336,28 +364,35 @@ namespace scribo //===== DEBUG ===== - - if (enable_debug) +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { // Bboxes image. - scribo::debug::save_bboxes_image(processed_image, lines, - scribo::make::debug_filename("step1_bboxes.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::bboxes_image(processed_image, lines), + "step1_bboxes"); // Bboxes enlarged - mln::io::ppm::save(scribo::debug::bboxes_enlarged_image(processed_image, lines), - scribo::make::debug_filename("step1_bboxes_enlarged.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::bboxes_enlarged_image(processed_image, lines), + "step1_bboxes_enlarged"); // Looks like a text line - mln::io::ppm::save(scribo::debug::looks_like_a_text_line_image(processed_image, lines), - scribo::make::debug_filename("step1_looks_like_a_text_line.ppm")); - + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::looks_like_a_text_line_image(processed_image, lines), + "step1_looks_like_a_text_line"); // mean and base lines. - mln::io::ppm::save(scribo::debug::mean_and_base_lines_image(processed_image, lines), - scribo::make::debug_filename("step1_x_height.ppm")); - + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::mean_and_base_lines_image(processed_image, lines), + "step1_x_height"); } +# endif // ! SCRIBO_NDEBUG //===== END OF DEBUG ===== @@ -366,23 +401,29 @@ namespace scribo //===== DEBUG ===== - if (enable_debug) +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { // mean and base lines. - mln::io::ppm::save(scribo::debug::mean_and_base_lines_image(processed_image, lines), - scribo::make::debug_filename("step2_x_height.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::mean_and_base_lines_image(processed_image, lines), + "step2_x_height"); // Looks like a text line - mln::io::ppm::save(scribo::debug::looks_like_a_text_line_image(processed_image, lines), - scribo::make::debug_filename("step2_looks_like_a_text_line.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::looks_like_a_text_line_image(processed_image, lines), + "step2_looks_like_a_text_line"); // Bboxes image. - scribo::debug::save_bboxes_image(processed_image, lines, - scribo::make::debug_filename("step2_bboxes.ppm")); - - + debug::logger().log_image( + debug::AuxiliaryResults, + scribo::debug::bboxes_image(processed_image, lines), + "step2_bboxes"); } +# endif // ! SCRIBO_NDEBUG //===== END OF DEBUG ===== on_progress(); @@ -404,20 +445,25 @@ namespace scribo //===== DEBUG ===== - if (enable_debug) +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { - image2d<value::rgb8> debug = data::convert(value::rgb8(), original_image); + image2d<value::rgb8> + debug = data::convert(value::rgb8(), original_image); for_all_lines(l, lines) { if (! lines(l).is_textline()) continue; mln::draw::box(debug, lines(l).bbox(), literal::blue); - mln::draw::line(debug, lines(l).bbox().pcenter(), lines(llinks(l)).bbox().pcenter(), literal::green); + mln::draw::line(debug, lines(l).bbox().pcenter(), + lines(llinks(l)).bbox().pcenter(), literal::green); } - mln::io::ppm::save(debug, scribo::make::debug_filename("links_raw.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + debug, "links_raw"); } +# endif // ! SCRIBO_NDEBUG //===== END OF DEBUG ===== on_progress(); @@ -428,22 +474,26 @@ namespace scribo llinks = scribo::filter::line_links_x_height(llinks); //===== DEBUG ===== - if (enable_debug) +# ifndef SCRIBO_NDEBUG + if (debug::logger().is_enabled()) { - image2d<value::rgb8> debug = data::convert(value::rgb8(), original_image); + image2d<value::rgb8> + debug = data::convert(value::rgb8(), original_image); for_all_links(i, llinks) if (llinks(i) && llinks(i) != i) mln::draw::line(debug, lines(i).bbox().pcenter(), lines(llinks(i)).bbox().pcenter(), literal::red); - mln::io::ppm::save(debug, scribo::make::debug_filename("links.ppm")); - + debug::logger().log_image(debug::AuxiliaryResults, + debug, "links"); for (unsigned i = 1; i < llinks.nelements(); ++i) llinks(i) = scribo::make::internal::find_root(llinks, i); debug = data::convert(value::rgb8(), original_image); - mln::util::array<accu::shape::bbox<point2d> > nbbox(llinks.nelements()); + mln::util::array<accu::shape::bbox<point2d> > + nbbox(llinks.nelements()); + for_all_lines(i, lines) { if (! lines(i).is_textline()) @@ -464,8 +514,10 @@ namespace scribo mln::draw::box(debug, b, literal::green); } - mln::io::ppm::save(debug, scribo::make::debug_filename("par.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + debug, "par"); } +# endif // ! SCRIBO_NDEBUG //===== END OF DEBUG ===== on_progress(); @@ -482,7 +534,7 @@ namespace scribo // Extract other Elements on_new_progress_label("Extracting Elements"); component_set<L> - elements = scribo::primitive::extract::non_text(doc, original_image); + elements = scribo::primitive::extract::non_text(doc, 3); on_progress(); diff --git a/scribo/scribo/toolchain/internal/text_in_doc_functor.hh b/scribo/scribo/toolchain/internal/text_in_doc_functor.hh index 16e981d..62074f0 100644 --- a/scribo/scribo/toolchain/internal/text_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/text_in_doc_functor.hh @@ -27,8 +27,6 @@ #ifndef SCRIBO_TOOLCHAIN_INTERNAL_TEXT_IN_DOC_FUNCTOR_HH # define SCRIBO_TOOLCHAIN_INTERNAL_TEXT_IN_DOC_FUNCTOR_HH -# include <mln/io/ppm/save.hh> - # include <scribo/core/def/lbl_type.hh> # include <scribo/primitive/extract/components.hh> @@ -54,8 +52,10 @@ # include <scribo/make/debug_filename.hh> -# include <scribo/debug/save_bboxes_image.hh> -# include <scribo/debug/save_linked_bboxes_image.hh> +# include <scribo/debug/logger.hh> +# include <scribo/debug/decision_image.hh> +# include <scribo/debug/bboxes_image.hh> +# include <scribo/debug/linked_bboxes_image.hh> # include <scribo/debug/bboxes_enlarged_image.hh> # include <scribo/debug/mean_and_base_lines_image.hh> # include <scribo/debug/looks_like_a_text_line_image.hh> @@ -77,7 +77,7 @@ namespace scribo struct text_in_doc_functor : public Toolchain_Functor { - typedef value::label<30> V; + typedef scribo::def::lbl_type V; typedef mln_ch_value(I,V) L; text_in_doc_functor(); @@ -97,7 +97,6 @@ namespace scribo bool enable_denoising; bool enable_line_seps; bool enable_whitespace_seps; - bool enable_debug; //============ // Parameters @@ -118,7 +117,6 @@ namespace scribo : enable_denoising(true), enable_line_seps(true), enable_whitespace_seps(true), - enable_debug(false), ocr_language("eng") { } @@ -164,19 +162,19 @@ namespace scribo on_progress(); } - if (enable_debug) + if (debug::logger().is_enabled()) { if (enable_whitespace_seps) - mln::io::pbm::save(whitespaces, - scribo::make::debug_filename("whitespaces.pbm")); + debug::logger().log_image(debug::AuxiliaryResults, + whitespaces, "whitespaces"); if (enable_line_seps) { - mln::io::pbm::save(separators, - scribo::make::debug_filename("vseparators.pbm")); + debug::logger().log_image(debug::AuxiliaryResults, + separators, "vseparators"); - mln::io::pbm::save(input_cleaned, - scribo::make::debug_filename("input_wo_vseparators.pbm")); + debug::logger().log_image(debug::AuxiliaryResults, + input_cleaned, "input_wo_vseparators"); } } @@ -187,9 +185,9 @@ namespace scribo input_cleaned = preprocessing::denoise_fg(input_cleaned, c8(), 3); - if (enable_debug) - mln::io::pbm::save(input_cleaned, - scribo::make::debug_filename("denoised.pbm")); + if (debug::logger().is_enabled()) + debug::logger().log_image(debug::AuxiliaryResults, + input_cleaned, "denoised"); on_progress(); } @@ -210,9 +208,9 @@ namespace scribo if (enable_whitespace_seps) components.add_separators(whitespaces); - if (enable_debug) - mln::io::pbm::save(components.separators(), - scribo::make::debug_filename("all_separators.pbm")); + if (debug::logger().is_enabled()) + debug::logger().log_image(debug::AuxiliaryResults, + components.separators(), "all_separators"); on_new_progress_label("Filtering components"); @@ -234,15 +232,19 @@ namespace scribo primitive::link::internal::dmax_width_and_height(1), anchor::MassCenter); - if (enable_debug) + if (debug::logger().is_enabled()) { - debug::save_linked_bboxes_image(input, left_link, right_link, - literal::blue, - literal::cyan, - literal::yellow, - literal::green, - anchor::MassCenter, - scribo::make::debug_filename("object_links.ppm")); + debug::logger().log_image( + debug::AuxiliaryResults, + debug::linked_bboxes_image(input, + left_link, + right_link, + literal::blue, + literal::cyan, + literal::yellow, + literal::green, + anchor::MassCenter), + "object_links"); } @@ -261,15 +263,16 @@ namespace scribo = filter::object_links_bbox_h_ratio(merged_links, 2.5f); - if (enable_debug) + if (debug::logger().is_enabled()) { mln_ch_value(I,value::rgb8) hratio_decision_image = scribo::debug::decision_image(input, merged_links, hratio_filtered_links, anchor::MassCenter); - io::ppm::save(hratio_decision_image, - scribo::make::debug_filename("hratio_links_decision_image.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + hratio_decision_image, + "hratio_links_decision_image"); } on_progress(); @@ -289,25 +292,29 @@ namespace scribo //===== DEBUG ===== - if (enable_debug) + if (debug::logger().is_enabled()) { // Bboxes image. - scribo::debug::save_bboxes_image(input, lines, - scribo::make::debug_filename("step1_bboxes.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::bboxes_image(input, lines), + "step1_bboxes"); // Bboxes enlarged - mln::io::ppm::save(scribo::debug::bboxes_enlarged_image(input, lines), - scribo::make::debug_filename("step1_bboxes_enlarged.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::bboxes_enlarged_image(input, lines), + "step1_bboxes_enlarged"); // Looks like a text line - mln::io::ppm::save(scribo::debug::looks_like_a_text_line_image(input, lines), - scribo::make::debug_filename("step1_looks_like_a_text_line.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::looks_like_a_text_line_image(input, lines), + "step1_looks_like_a_text_line"); // mean and base lines. - mln::io::ppm::save(scribo::debug::mean_and_base_lines_image(input, lines), - scribo::make::debug_filename("step1_x_height.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::mean_and_base_lines_image(input, lines), + "step1_x_height"); } //===== END OF DEBUG ===== @@ -320,27 +327,29 @@ namespace scribo //===== DEBUG ===== - if (enable_debug) + if (debug::logger().is_enabled()) { // mean and base lines. - mln::io::ppm::save(scribo::debug::mean_and_base_lines_image(input, lines), - scribo::make::debug_filename("step2_x_height.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::mean_and_base_lines_image(input, lines), + "step2_x_height"); // Looks like a text line - mln::io::ppm::save(scribo::debug::looks_like_a_text_line_image(input, lines), - scribo::make::debug_filename("step2_looks_like_a_text_line.ppm")); + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::looks_like_a_text_line_image(input, lines), + "step2_looks_like_a_text_line"); // Bboxes image. - scribo::debug::save_bboxes_image(input, lines, - scribo::make::debug_filename("step2_bboxes.ppm")); - + debug::logger().log_image(debug::AuxiliaryResults, + scribo::debug::bboxes_image(input, lines), + "step2_bboxes"); } - if (enable_debug) + if (debug::logger().is_enabled()) { std::ofstream file(scribo::make::debug_filename("step2_bboxes_100p.txt").c_str()); diff --git a/scribo/scribo/toolchain/nepomuk/text_extraction.hh b/scribo/scribo/toolchain/nepomuk/text_extraction.hh index 0447cef..2534ce8 100644 --- a/scribo/scribo/toolchain/nepomuk/text_extraction.hh +++ b/scribo/scribo/toolchain/nepomuk/text_extraction.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. // @@ -121,8 +122,7 @@ namespace scribo // Run document toolchain. lines_bg = scribo::toolchain::text_in_doc(input_bin, true, - language.toUtf8().data(), - false); + language.toUtf8().data()); // Negate document. logical::not_inplace(input_bin); @@ -130,8 +130,7 @@ namespace scribo // Run document toolchain. lines_fg = scribo::toolchain::text_in_doc(input_bin, true, - language.toUtf8().data(), - false); + language.toUtf8().data()); } diff --git a/scribo/scribo/toolchain/text_in_doc.hh b/scribo/scribo/toolchain/text_in_doc.hh index e6ba69e..9f20b39 100644 --- a/scribo/scribo/toolchain/text_in_doc.hh +++ b/scribo/scribo/toolchain/text_in_doc.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. // @@ -47,8 +47,7 @@ namespace scribo text_in_doc(const Image<I>& input, bool denoise, const std::string& language = std::string("eng"), bool find_line_seps = true, - bool find_whitespace_seps = true, - bool debug = false); + bool find_whitespace_seps = true); # ifndef MLN_INCLUDE_ONLY @@ -59,14 +58,12 @@ namespace scribo text_in_doc(const Image<I>& input, bool denoise, const std::string& language = std::string("eng"), bool find_line_seps = true, - bool find_whitespace_seps = true, - bool debug = false) + bool find_whitespace_seps = true) { internal::text_in_doc_functor<I> f; f.enable_denoising = denoise; f.enable_line_seps = find_line_seps; f.enable_whitespace_seps = find_whitespace_seps; - f.enable_debug = debug; f.ocr_language = language; line_set<mln_ch_value(I, def::lbl_type)> lines = f(input); diff --git a/scribo/src/content_in_doc.cc b/scribo/src/content_in_doc.cc index d8d4e52..81ec4fb 100644 --- a/scribo/src/content_in_doc.cc +++ b/scribo/src/content_in_doc.cc @@ -40,8 +40,7 @@ #include <scribo/core/document.hh> #include <scribo/debug/usage.hh> - -#include <scribo/make/debug_filename.hh> +#include <scribo/debug/logger.hh> #include <scribo/preprocessing/crop_without_localization.hh> #include <scribo/preprocessing/crop.hh> @@ -79,13 +78,12 @@ int main(int argc, char* argv[]) "input.* out.xml <denoise_enabled> [<pmin_row> <pmin_col> <pmax_row> <pmax_col>] [language] [find_lines] [find_whitespaces] [K] [debug_dir]", args_desc); - bool debug = false; - // Enable debug output. if (argc == 9 || argc == 13) { + scribo::debug::logger().set_filename_prefix(argv[argc - 1]); + scribo::debug::logger().set_level(scribo::debug::All); scribo::make::internal::debug_filename_prefix = argv[argc - 1]; - debug = true; } trace::entering("main"); @@ -109,7 +107,6 @@ int main(int argc, char* argv[]) std::cout << "Using K = " << K << std::endl; } - image2d<bool> tmp_fg; input_preproc = toolchain::text_in_doc_preprocess(input, false, K); } @@ -130,9 +127,8 @@ int main(int argc, char* argv[]) input_preproc = preprocessing::crop_without_localization(input_preproc, roi); crop_shift = point2d(minr, minc); - if (debug) - mln::io::pbm::save(input_preproc, - scribo::make::debug_filename("input_preproc_cropped.pbm")); + scribo::debug::logger().log_image(scribo::debug::Results, input_preproc, + "input_preproc_cropped.pbm"); } bool denoise = (argc > 3 && atoi(argv[3]) != 0); @@ -159,7 +155,7 @@ int main(int argc, char* argv[]) << " ocr_language = " << language << " | find_lines_seps = " << find_line_seps << " | find_whitespace_seps = " << find_whitespace_seps - << " | debug = " << debug + << " | debug = " << scribo::debug::logger().is_enabled() << std::endl; // Run document toolchain. @@ -167,11 +163,12 @@ int main(int argc, char* argv[]) // Text std::cout << "Analysing document..." << std::endl; document<L> - doc = scribo::toolchain::content_in_doc(input, input_preproc, denoise, language, + doc = scribo::toolchain::content_in_doc(input, input_preproc, denoise, find_line_seps, find_whitespace_seps, - debug); + !language.empty(), language); // Saving results + std::cout << "Saving results..." << std::endl; scribo::io::xml::save(doc, argv[2], scribo::io::xml::PageExtended); scribo::io::xml::save(doc, "page.xml", scribo::io::xml::Page); scribo::io::xml::save(doc, "full.xml", scribo::io::xml::Full); diff --git a/scribo/src/pbm_text_in_doc.cc b/scribo/src/pbm_text_in_doc.cc index 2726ead..42b7d88 100644 --- a/scribo/src/pbm_text_in_doc.cc +++ b/scribo/src/pbm_text_in_doc.cc @@ -45,8 +45,6 @@ #include <scribo/debug/usage.hh> -#include <scribo/make/debug_filename.hh> - #include <scribo/preprocessing/crop_without_localization.hh> #include <scribo/io/text_boxes/save.hh> @@ -83,13 +81,12 @@ int main(int argc, char* argv[]) "input.pbm out.txt <denoise_enabled> [<pmin_row> <pmin_col> <pmax_row> <pmax_col>] <language> <find_lines> <find_whitespaces> <debug_dir>", args_desc); - bool debug = false; - // Enable debug output. if (argc == 8 || argc == 12) { + scribo::debug::logger().set_filename_prefix(argv[argc - 1]); + scribo::debug::logger().set_level(scribo::debug::All); scribo::make::internal::debug_filename_prefix = argv[argc - 1]; - debug = true; } trace::entering("main"); @@ -116,9 +113,8 @@ int main(int argc, char* argv[]) input = preprocessing::crop_without_localization(input, roi); crop_shift = point2d(minr, minc); - if (debug) - mln::io::pbm::save(input, - scribo::make::debug_filename("input_cropped.pbm")); + scribo::debug::logger().log_image(scribo::debug::Results, input, + "input_cropped.pbm"); } bool denoise = (argc > 3 && atoi(argv[3]) != 0); @@ -146,14 +142,14 @@ int main(int argc, char* argv[]) << " ocr language = " << language << " | find_lines_seps = " << find_line_seps << " | find_whitespace_seps = " << find_whitespace_seps - << " | debug = " << debug + << " | debug = " << scribo::debug::logger().is_enabled() << std::endl; // Run document toolchain. line_set<L> lines = scribo::toolchain::text_in_doc(input, denoise, language, find_line_seps, - find_whitespace_seps, debug); + find_whitespace_seps); scribo::document<L> doc; doc.set_filename(argv[1]); diff --git a/scribo/src/primitive/extract/separators_nonvisible.cc b/scribo/src/primitive/extract/separators_nonvisible.cc index d50cc73..177047c 100644 --- a/scribo/src/primitive/extract/separators_nonvisible.cc +++ b/scribo/src/primitive/extract/separators_nonvisible.cc @@ -34,6 +34,7 @@ const char *args_desc[][2] = { { "input.pbm", "A binary image." }, { "output.pbm", "Output image." }, + { "enable_debug", "0 or 1 (default 0)" }, {0, 0} }; @@ -43,10 +44,10 @@ int main(int argc, char *argv[]) using namespace mln; using namespace scribo; - if (argc != 3) + if (argc != 3 && argc != 4) return scribo::debug::usage(argv, "Extract non visible separators (whitespaces)", - "input.pbm output.pbm", + "input.pbm output.pbm [enable_debug]", args_desc); trace::entering("main"); @@ -54,6 +55,9 @@ int main(int argc, char *argv[]) image2d<bool> input; io::pbm::load(input, argv[1]); + if (argc > 3 && atoi(argv[3])) + scribo::debug::logger().set_level(scribo::debug::All); + io::pbm::save(primitive::extract::separators_nonvisible(input), argv[2]); trace::exiting("main"); -- 1.5.6.5