[Olena-patches] last-svn-commit-112-g3d900b5 Prepare DIA tools for Nuxeo/XWiki.

* scribo/src/Makefile.am: Add a new target. * scribo/src/text_in_article_preprocess.cc: New. * scribo/src/text_in_article_pbm.cc: Add optional cropping and makes the debug optionnal. * scribo/text/clean.hh: Improve cleanup. * scribo/text/recognition.hh: Remove last '\n' in Tesseract's output. --- scribo/ChangeLog | 16 ++ scribo/src/Makefile.am | 9 + scribo/src/text_in_article_pbm.cc | 162 +++++++++++++------- ...sauvola_ms.cc => text_in_article_preprocess.cc} | 90 +++++------- scribo/text/clean.hh | 13 +- scribo/text/recognition.hh | 14 ++- 6 files changed, 184 insertions(+), 120 deletions(-) copy scribo/src/{binarization/fg_sauvola_ms.cc => text_in_article_preprocess.cc} (53%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index f2850e0..59f9e67 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,19 @@ +2010-05-04 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; + + Prepare DIA tools for Nuxeo/XWiki. + + * scribo/src/Makefile.am: Add a new target. + + * scribo/src/text_in_article_preprocess.cc: New. + + * scribo/src/text_in_article_pbm.cc: Add optional cropping and + makes the debug optionnal. + + * scribo/text/clean.hh: Improve cleanup. + + * scribo/text/recognition.hh: Remove last '\n' in Tesseract's + output. + 2010-04-30 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; Improve OCR recognition. diff --git a/scribo/src/Makefile.am b/scribo/src/Makefile.am index 80dc782..08b4bd9 100644 --- a/scribo/src/Makefile.am +++ b/scribo/src/Makefile.am @@ -82,6 +82,15 @@ if HAVE_TIFF text_in_article_pbm_SOURCES = text_in_article_pbm.cc + bin_PROGRAMS += text_in_article_preprocess + text_in_article_preprocess_CPPFLAGS = $(AM_CPPFLAGS) \ + `Magick++-config --cppflags` + text_in_article_preprocess_LDFLAGS = $(AM_LDFLAGS) \ + -lpthread `Magick++-config --libs` + text_in_article_preprocess_SOURCES = text_in_article_preprocess.cc + + + bin_PROGRAMS += text_in_photo_fast text_in_photo_fast_SOURCES = text_in_photo_fast.cc text_in_photo_fast_CPPFLAGS = $(AM_CPPFLAGS) \ diff --git a/scribo/src/text_in_article_pbm.cc b/scribo/src/text_in_article_pbm.cc index e71d21b..20a18e9 100644 --- a/scribo/src/text_in_article_pbm.cc +++ b/scribo/src/text_in_article_pbm.cc @@ -40,6 +40,8 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> +#include <mln/data/paste_without_localization.hh> + #include <scribo/core/line_set.hh> #include <scribo/primitive/extract/components.hh> @@ -85,6 +87,10 @@ const char *args_desc[][2] = for the background." }, { "out.txt", "Text output" }, { "denoise", "1 enables denoising, 0 disables it. (enabled by default)" }, + { "pmin_row", "Row index of the top left corner of the Region of interest." }, + { "pmin_col", "Col index of the top left corner of the Region of interest." }, + { "pmax_row", "Row index of the bottom right corner of the Region of interest." }, + { "pmax_col", "Col index of the bottom right corner of the Region of interest." }, { "debug_dir", "Output directory for debug image" }, {0, 0} }; @@ -95,14 +101,20 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 3 && argc != 4 && argc != 5) + if (argc != 3 && argc != 4 && argc != 5 && argc != 8 && argc != 9) return scribo::debug::usage(argv, "Find text lines using left/right validation and display x-height in a binarized article.", - "input.pbm out.txt <denoise: 0|1> <debug_dir>", + "input.pbm out.txt <denoise: 0|1> [<pmin_row> <pmin_col> <pmax_row> <pmax_col>] <debug_dir>", args_desc); - if (argc == 5) - scribo::make::internal::debug_filename_prefix = argv[4]; + bool debug = false; + + // Enable debug output. + if (argc == 5 || argc == 9) + { + scribo::make::internal::debug_filename_prefix = argv[argc - 1]; + debug = true; + } trace::entering("main"); @@ -110,6 +122,27 @@ int main(int argc, char* argv[]) image2d<bool> input; mln::io::pbm::load(input, argv[1]); + + // Optional Cropping + if (argc >= 8) + { + def::coord + minr = atoi(argv[4]), + minc = atoi(argv[5]), + maxr = atoi(argv[6]), + maxc = atoi(argv[7]); + + box2d roi = mln::make::box2d(minr, minc, maxr, maxc); + image2d<bool> tmp(maxr - minr + 1, maxc - minc + 1); + data::paste_without_localization(input | roi, tmp); + input = tmp; + + if (debug) + mln::io::pbm::save(input, + scribo::make::debug_filename("input_cropped.pbm")); + } + + typedef value::label_16 V; typedef image2d<V> L; @@ -129,7 +162,9 @@ int main(int argc, char* argv[]) // whitespaces += separators; - mln::io::pbm::save(separators, "vseparators.pbm"); + if (debug) + mln::io::pbm::save(separators, + scribo::make::debug_filename("vseparators.pbm")); // mln::io::pbm::save(whitespaces, "separators.pbm"); // mln::io::pbm::save(input_cleaned, "input_no_separators.pbm"); @@ -214,18 +249,20 @@ int main(int argc, char* argv[]) //===== DEBUG ===== - // Bboxes image. - scribo::debug::save_bboxes_image(input, lines, - scribo::make::debug_filename("step1_bboxes.ppm")); + if (debug) + { - // Bboxes enlarged - mln::io::ppm::save(scribo::debug::bboxes_enlarged_image(input, lines), - scribo::make::debug_filename("step1_bboxes_enlarged.ppm")); + // Bboxes image. + scribo::debug::save_bboxes_image(input, lines, + scribo::make::debug_filename("step1_bboxes.ppm")); - // 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")); + // Bboxes enlarged + mln::io::ppm::save(scribo::debug::bboxes_enlarged_image(input, lines), + scribo::make::debug_filename("step1_bboxes_enlarged.ppm")); + // 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")); // // Bboxes + line infos // { @@ -261,10 +298,11 @@ int main(int argc, char* argv[]) // } - // 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")); + // 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")); + } //===== END OF DEBUG ===== @@ -277,56 +315,62 @@ int main(int argc, char* argv[]) //===== DEBUG ===== - // 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")); + if (debug) + { - // 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")); + // 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")); - // Bboxes image. - scribo::debug::save_bboxes_image(input, lines, - scribo::make::debug_filename("step2_bboxes.ppm")); + // 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")); + // Bboxes image. + scribo::debug::save_bboxes_image(input, lines, + scribo::make::debug_filename("step2_bboxes.ppm")); - { - std::ofstream file(scribo::make::debug_filename("step2_bboxes_100p.txt").c_str()); -// std::ofstream file_50p(scribo::make::debug_filename("step2_bboxes_50p.txt").c_str()); - - for_all_lines(l, lines) - if (lines(l).tag() != line::Merged - && lines(l).tag() != line::Ignored - && lines(l).tag() != line::Pathological) - { - file << lines(l).bbox().pmin().row() << " " - << lines(l).bbox().pmin().col() << " " - << lines(l).bbox().pmax().row() << " " - << lines(l).bbox().pmax().col() << " " - << lines(l).card() << " " - << lines(l).baseline() << " " - << lines(l).x_height() << " " - << lines(l).meanline() << " " - << lines(l).d_height() << " " - << lines(l).a_height() << " " - << lines(l).char_space() << " " - << lines(l).char_width() << std::endl; - -// file_50p << lines(l).bbox().pmin().row() / 2 << " " -// << lines(l).bbox().pmin().col() / 2 << " " -// << lines(l).bbox().pmax().row() / 2 << " " -// << lines(l).bbox().pmax().col() / 2 << std::endl; - } - - file.close(); -// file_50p.close(); - } + } - //===== END OF DEBUG ===== + +// { +// std::ofstream file(scribo::make::debug_filename("step2_bboxes_100p.txt").c_str()); +// // std::ofstream file_50p(scribo::make::debug_filename("step2_bboxes_50p.txt").c_str()); + +// for_all_lines(l, lines) +// if (lines(l).tag() != line::Merged +// && lines(l).tag() != line::Ignored +// && lines(l).tag() != line::Pathological) +// { +// file << lines(l).bbox().pmin().row() << " " +// << lines(l).bbox().pmin().col() << " " +// << lines(l).bbox().pmax().row() << " " +// << lines(l).bbox().pmax().col() << " " +// << lines(l).card() << " " +// << lines(l).baseline() << " " +// << lines(l).x_height() << " " +// << lines(l).meanline() << " " +// << lines(l).d_height() << " " +// << lines(l).a_height() << " " +// << lines(l).char_space() << " " +// << lines(l).char_width() << std::endl; + +// // file_50p << lines(l).bbox().pmin().row() / 2 << " " +// // << lines(l).bbox().pmin().col() / 2 << " " +// // << lines(l).bbox().pmax().row() / 2 << " " +// // << lines(l).bbox().pmax().col() / 2 << std::endl; +// } + +// file.close(); +// // file_50p.close(); +// } + + + //===== END OF DEBUG ===== - scribo::io::xml::save_text_lines(argv[1], lines, "out.xml"); + scribo::io::xml::save_text_lines(argv[1], lines, "out.xml"); diff --git a/scribo/src/binarization/fg_sauvola_ms.cc b/scribo/src/text_in_article_preprocess.cc similarity index 53% copy from scribo/src/binarization/fg_sauvola_ms.cc copy to scribo/src/text_in_article_preprocess.cc index 9c606d6..8cfa82f 100644 --- a/scribo/src/binarization/fg_sauvola_ms.cc +++ b/scribo/src/text_in_article_preprocess.cc @@ -1,5 +1,4 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2010 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -24,82 +23,69 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. + +#include <libgen.h> +#include <iostream> + #include <mln/core/image/image2d.hh> -#include <mln/value/rgb8.hh> + #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> -#include <scribo/binarization/sauvola_ms.hh> -#include <scribo/preprocessing/split_bg_fg.hh> -#include <scribo/debug/usage.hh> +#include <mln/value/rgb8.hh> -bool check_args(int argc, char * argv[]) -{ - if (argc != 7) - return false; +#include <mln/logical/not.hh> - int s = atoi(argv[4]); +#include <scribo/binarization/sauvola_ms.hh> - if (s < 2 || s > 3) - { - std::cout << "s must be set to 2 or 3." - << std::endl; - return false; - } +#include <scribo/debug/usage.hh> + +#include <scribo/preprocessing/split_bg_fg.hh> - return true; -} const char *args_desc[][2] = { { "input.*", "An image." }, - { "output.pbm", "A binary image." }, - { "lambda", "Lambda used to split bg/fg." }, - { "w", "Window size at scale 1. (Common value: 101)" }, - { "s", "First subsampling ratio (Common value: 3)." }, - { "min_area", "Minimum object area at scale 1 (Common value: 67)" }, + { "output.pbm", "A text file with all the recognized text" }, + { "enable fg/bg", "If set to 1 enables foreground extraction. (disabled by default)" }, + { "lambda", "Lambda used in remove fg/bg (Automaticaly deduced by default)." }, {0, 0} }; - - -int main(int argc, char *argv[]) +int main(int argc, char* argv[]) { - using namespace mln; using namespace scribo; + using namespace mln; - if (!check_args(argc, argv)) + if (argc != 3 && argc != 4 && argc != 5) return scribo::debug::usage(argv, - "Multi-Scale Binarization based on Sauvola's algorithm. Performs a binarization on each component of the color image and merges the results.", - "input.* output.pbm w s area_threshold", + "Find text in a color document.", + "input.* output.pbm <enable fg/bg> <lambda>", args_desc); - trace::entering("main"); - - unsigned lambda = atoi(argv[2]); + image2d<value::rgb8> input_rgb; + io::magick::load(input_rgb, argv[1]); - // Window size - unsigned w_1 = atoi(argv[3]); // Scale 1 + unsigned lambda; + if (argc == 5) + lambda = atoi(argv[4]); + else + lambda = 1.2 * (input_rgb.nrows() + input_rgb.ncols()); - // First subsampling scale. - unsigned s = atoi(argv[4]); - - // Lambda value - unsigned lambda_min_1 = atoi(argv[5]); - - - image2d<value::rgb8> input_1; - io::magick::load(input_1, argv[1]); + // Extract foreground + if (argc == 4 && atoi(argv[3]) == 1) + { + std::cout << "Extracting foreground..." << std::endl; + input_rgb = preprocessing::split_bg_fg(input_rgb, lambda, 32).second(); + } - image2d<value::rgb8> - fg = scribo::preprocessing::split_bg_fg(input_1, lambda, 32).first(); + // Binarize foreground to use it in the processing chain. + std::cout << "Binarizing foreground..." << std::endl; + image2d<bool> input = scribo::binarization::sauvola_ms(input_rgb, 101, 3); - image2d<bool> - output = scribo::binarization::sauvola_ms(fg, w_1, s, lambda_min_1, SCRIBO_DEFAULT_SAUVOLA_K); + logical::not_inplace(input); - io::pbm::save(output, argv[6]); + mln::io::pbm::save(input, argv[2]); } - - diff --git a/scribo/text/clean.hh b/scribo/text/clean.hh index fa31ebe..f93bc6f 100644 --- a/scribo/text/clean.hh +++ b/scribo/text/clean.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of Olena. // @@ -113,14 +114,14 @@ namespace scribo if (fact < 1) { std::cout << "Upsampling..." << " - " - << std::ceil(fact) << std::endl; - while (fact < 1) + << fact << std::endl; + while (fact < 0.90) { output = scribo::upsampling::bs2x(output); // 2x upsampling fact *= 2.0f; -// std::cout << "fact = " << fact -// << " - output.domain = " << output.domain() -// << std::endl; +// std::cout << "fact = " << fact +// << " - output.domain = " << output.domain() +// << std::endl; } } else if (fact > 2.5f) diff --git a/scribo/text/recognition.hh b/scribo/text/recognition.hh index 72df1e2..6f9dac7 100644 --- a/scribo/text/recognition.hh +++ b/scribo/text/recognition.hh @@ -159,7 +159,7 @@ namespace scribo /// text_ima_cleaned domain is larger than text_ima's. I text_ima_cleaned = text::clean(lines(i), text_ima); - mln::io::pbm::save(text_ima_cleaned, mln::debug::filename("line.pbm", debug_id++)); +// mln::io::pbm::save(text_ima_cleaned, mln::debug::filename("line.pbm", debug_id++)); // Setting objects to 'True' logical::not_inplace(text_ima_cleaned); @@ -182,7 +182,11 @@ namespace scribo { std::cerr << s << std::endl; if (output_file != 0) - file << lines(i).bbox() << " " << s << std::endl; + { + std::string str(s); + str = str.substr(0, str.length() - 1); + file << lines(i).bbox() << " " << str; + } } // The string has been allocated by Tesseract. We must free it. @@ -241,7 +245,11 @@ namespace scribo { std::cout << s << std::endl; if (output_file != 0) - file << line.domain() << " " << s << std::endl; + { + std::string str(s); + str = str.substr(0, str.length() - 1); + file << line.domain() << " " << str; + } } // The string has been allocated by Tesseract. We must free it. -- 1.5.6.5