[Olena-patches] olena-2.0-97-gf9ea569 Cleanup Kim's binarization.

* src/binarization/kim.cc: Move code... * scribo/binarization/kim.hh: ... here. * tests/binarization/Makefile.am: Add new target. * tests/binarization/kim.cc, * tests/binarization/kim.ref.pbm, * tests/binarization/kim_weven_hodd.ref.pbm, * tests/binarization/kim_wodd_heven.ref.pbm, * tests/binarization/kim_wodd_hodd.ref.pbm: New. --- scribo/scribo/binarization/kim.hh | 262 ++++++++++++++++++++++ scribo/src/binarization/kim.cc | 199 +--------------- scribo/tests/binarization/Makefile.am | 6 + scribo/tests/binarization/{sauvola.cc => kim.cc} | 18 +- scribo/tests/binarization/kim.ref.pbm | Bin 0 -> 32884 bytes scribo/tests/binarization/kim_weven_hodd.ref.pbm | Bin 0 -> 32820 bytes scribo/tests/binarization/kim_wodd_heven.ref.pbm | Bin 0 -> 32884 bytes scribo/tests/binarization/kim_wodd_hodd.ref.pbm | Bin 0 -> 32820 bytes 8 files changed, 289 insertions(+), 196 deletions(-) create mode 100644 scribo/scribo/binarization/kim.hh copy scribo/tests/binarization/{sauvola.cc => kim.cc} (77%) create mode 100644 scribo/tests/binarization/kim.ref.pbm create mode 100644 scribo/tests/binarization/kim_weven_hodd.ref.pbm create mode 100644 scribo/tests/binarization/kim_wodd_heven.ref.pbm create mode 100644 scribo/tests/binarization/kim_wodd_hodd.ref.pbm diff --git a/scribo/scribo/binarization/kim.hh b/scribo/scribo/binarization/kim.hh new file mode 100644 index 0000000..0d31a1d --- /dev/null +++ b/scribo/scribo/binarization/kim.hh @@ -0,0 +1,262 @@ +// Copyright (C) 2012 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_BINARIZATION_KIM_HH +# define SCRIBO_BINARIZATION_KIM_HH + +/// \file +/// +/// + +#include <cmath> +#include <map> +#include <scribo/binarization/sauvola.hh> +#include <scribo/util/integral_sum_sum2_functor.hh> +#include <scribo/text/extract_lines.hh> + +namespace scribo +{ + + namespace binarization + { + + using namespace mln; + + /*! \brief Kim's binarization + + This algorithms performs a first rough binarization on the input + (here we use Sauvola's method). Then text lines are roughly + detected by grouping connected components. For each text lines, + character thickness and x height is computed. Finally, each + lines is binarized again, using text features as parameters for + adjusting the final threshold. + + This algorithms considers that global and local statistics + should be used to compute a threshold in text areas. + + This is an improvement of Sauvola's method. + + \param[in] input A gray-level image. + \param[in] window_size the window size to be used for the first + binarization. + \param[in] k Sauvola's formula parameter. + + This implementation is based on the paper "Multi-Window + Binarization of Camera Image for Document Recognition", In-Jung + Kim, Proceedings of the 9th International Workshop on Frontiers + in Handwriting Recognition (IWFHR-9 2004) + */ + template <typename I> + mln_ch_value(I, bool) + kim(const Image<I>& input, unsigned window_size, double k); + + /*! \overload + k is set to SCRIBO_DEFAULT_SAUVOLA_K. + */ + template <typename I> + mln_ch_value(I, bool) + kim(const Image<I>& input, unsigned window_size); + + +# ifndef MLN_INCLUDE_ONLY + + namespace internal + { + + double + compute_thres(const mln::image2d<mln::util::couple<double,double> >& integral_sum_sum_2, + int row, int col, unsigned win_size, + const scribo::binarization::internal::sauvola_formula& formula) + { + point2d + tl(row - win_size - 1, + col - win_size - 1), + br(row + win_size, + col + win_size); + + box2d b(tl, br); + b.crop_wrt(integral_sum_sum_2.domain()); + + point2d tr = b.pmax(); + tr.row() = b.pmin().row(); + point2d bl = b.pmin(); + bl.row() = b.pmax().row(); + + unsigned card_min = b.nsites() - b.height() - b.width() + 1; + + const mln::util::couple<double,double>& + D = integral_sum_sum_2(b.pmax()), + B = integral_sum_sum_2(tr), + C = integral_sum_sum_2(bl), + A = integral_sum_sum_2(b.pmin()); + + double sum = D.first() - B.first() - C.first() + A.first(); + double sum_2 = D.second() - B.second() - C.second() + A.second(); + double mean = sum / card_min; + + double num = (sum_2 - sum * sum / card_min); + double stddev; + if (num > 0) + stddev = std::sqrt(num / (card_min - 1)); + else + stddev = 0; + + return formula(mean, stddev); + } + + } // end of namespace scribo::binarization::internal + + + template <typename I> + mln_ch_value(I, bool) + kim(const Image<I>& input_, unsigned window_size, double k) + { + trace::entering("scribo::binarization::kim"); + + const I& input = exact(input_); + + mln_precondition(input.is_valid()); + + // 1st simple binarization + image2d<bool> + output = scribo::binarization::sauvola(input, window_size, k); + + // Compute integral image + scribo::util::integral_sum_sum2_functor<mln_value(I),double> f_sum_sum2; + image2d<mln::util::couple<double,double> > + integral_sum_sum_2 = scribo::util::init_integral_image(input, 1, f_sum_sum2); + + // Find text lines + line_set<image2d<scribo::def::lbl_type> > + lines = scribo::text::extract_lines(output, c8()); + + typedef scribo::def::lbl_type V; + typedef image2d<V> L; + mln::util::array<unsigned> thickness(lines.nelements() + 1, 0); + const component_set<L>& comp_set = lines.components(); + const L& lbl = comp_set.labeled_image(); + + // Compute run-lengths histogram in order to compute character + // thickness for each line. + for_all_lines(i, lines) + { + if (!lines(i).is_textline()) + continue; + + std::map<unsigned, unsigned> histo; + int count = 0; + for (int l = lines(i).bbox().pmin().row(); l <= lines(i).bbox().pmax().row(); ++l) + { + const V* end_ptr = &lbl.at_(l, lines(i).bbox().pmax().col() + 1); + for (const V* run_ptr = &lbl.at_(l, lines(i).bbox().pmin().col()); run_ptr != end_ptr; ++run_ptr) + if (*run_ptr) + ++count; + else + if (count) + { + if (histo.find(count) != histo.end()) + histo[count]++; + else + histo.insert(std::make_pair(count, 1)); + count = 0; + } + } + + unsigned max = 0; + unsigned thick = 0; + for (std::map<unsigned, unsigned>::const_iterator it = histo.begin(); it != histo.end(); ++it) + if (it->second > max) + { + max = it->second; + thick = it->first; + } + + thickness(lines(i).id()) = thick; + } + + // Compute thresholds for each pixel of each line and binarize again! + for_all_lines(i, lines) + { + if (!lines(i).is_textline()) + continue; + + double + win_min = thickness(lines(i).id()), + win_max = lines(i).bbox().height();; + + mln_assertion(win_min != 0); + mln_assertion(win_max != 0); + + scribo::binarization::internal::sauvola_formula formula; + + double teta = 0.3; // Good results from 0.1 to 0.3 according + // to the paper. + for (int row = lines(i).bbox().pmin().row(); + row <= lines(i).bbox().pmax().row(); + ++row) + { + bool* out_ptr = &output.at_(row, lines(i).bbox().pmin().col()); + const mln_value(I)* in_ptr = &input.at_(row, lines(i).bbox().pmin().col()); + for (int col = lines(i).bbox().pmin().col(); + col <= lines(i).bbox().pmax().col(); + ++col) + { + // Min case + double T_min = compute_thres(integral_sum_sum_2, row, col, win_min, formula); + + // Max case + double T_max = compute_thres(integral_sum_sum_2, row, col, win_max, formula); + + // Final threshold + double T = teta * T_max + (1 - teta) * T_min; + + mln_assertion(T_min <= 255); + mln_assertion(T_max <= 255); + mln_assertion(T <= 255); + + *out_ptr++ = *in_ptr++ <= T; + } + } + } + + trace::exiting("scribo::binarization::kim"); + return output; + } + + + template <typename I> + mln_ch_value(I, bool) + kim(const Image<I>& input, unsigned window_size) + { + return kim(input, window_size, SCRIBO_DEFAULT_SAUVOLA_K); + } + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::binarization + +} // end of namespace scribo + +#endif // ! SCRIBO_BINARIZATION_KIM_HH diff --git a/scribo/src/binarization/kim.cc b/scribo/src/binarization/kim.cc index aa81f50..9f9c38f 100644 --- a/scribo/src/binarization/kim.cc +++ b/scribo/src/binarization/kim.cc @@ -28,26 +28,18 @@ #include <cmath> #include <mln/core/image/image2d.hh> #include <mln/value/int_u8.hh> -#include <mln/io/magick/load.hh> -#include <mln/io/pbm/save.hh> +#include <mln/io/magick/all.hh> #include <mln/data/transform.hh> #include <mln/fun/v2v/rgb_to_luma.hh> - -#include <scribo/binarization/sauvola_ms.hh> +#include <scribo/binarization/kim.hh> #include <scribo/debug/option_parser.hh> #include <scribo/debug/logger.hh> -#include <scribo/util/integral_sum_sum2_functor.hh> - -#include <scribo/text/extract_lines.hh> -#include <scribo/debug/bboxes_enlarged_image.hh> -#include <algorithm> - static const scribo::debug::arg_data arg_desc[] = { { "input.*", "An image." }, - { "output.pbm", "A binary image." }, + { "output.*", "A binary image." }, {0, 0} }; @@ -67,59 +59,11 @@ static const scribo::debug::opt_data opt_desc[] = { "debug-prefix", "Enable debug image outputs. Prefix image name with that " "given prefix.", "<prefix>", 0, 1, 0 }, { "k", "Sauvola's formulae parameter", "<value>", 0, 1, "0.34" }, - { "s", "First subsampling ratio. Possible values: 2 or 3.", "ratio", - scribo::debug::check_sauvola_first_subsampling, 1, "3" }, { "verbose", "Enable verbose mode", 0, 0, 0, 0 }, { "win-size", "Window size at scale 1", "<size>", 0, 1, "101" }, {0, 0, 0, 0, 0, 0} }; - - - -double -compute_thres(const mln::image2d<mln::util::couple<double,double> >& integral_sum_sum_2, - int row, int col, unsigned win_size, - const scribo::binarization::internal::sauvola_formula& formula) -{ - point2d - tl(row - win_size - 1, - col - win_size - 1), - br(row + win_size, - col + win_size); - - box2d b(tl, br); - b.crop_wrt(integral_sum_sum_2.domain()); - - point2d tr = b.pmax(); - tr.row() = b.pmin().row(); - point2d bl = b.pmin(); - bl.row() = b.pmax().row(); - - unsigned card_min = b.nsites() - b.height() - b.width() + 1; - - const mln::util::couple<double,double>& - D = integral_sum_sum_2(b.pmax()), - B = integral_sum_sum_2(tr), - C = integral_sum_sum_2(bl), - A = integral_sum_sum_2(b.pmin()); - - double sum = D.first() - B.first() - C.first() + A.first(); - double sum_2 = D.second() - B.second() - C.second() + A.second(); - double mean = sum / card_min; - - double num = (sum_2 - sum * sum / card_min); - double stddev; - if (num > 0) - stddev = std::sqrt(num / (card_min - 1)); - else - stddev = 0; - - return formula(mean, stddev); -} - - - int main(int argc, char *argv[]) { using namespace mln; @@ -142,16 +86,17 @@ int main(int argc, char *argv[]) trace::entering("main"); - bool verbose = options.is_set("verbose"); // Window size unsigned w_1 = atoi(options.opt_value("win-size").c_str()); - // First subsampling scale. - unsigned s = atoi(options.opt_value("s").c_str()); double k = atof(options.opt_value("k").c_str()); - if (verbose) - std::cout << "Using w_1=" << w_1 << " - s=" << s - << " - k=" << k << std::endl; + if (options.is_set("verbose")) + { + scribo::debug::logger().set_verbose_mode( + scribo::debug::txt_to_verbose_mode(options.opt_value("verbose"))); + scribo::debug::logger().log(Low, std::string("Using w_1=") + w_1 + + std::string(" - k=") + k); + } // Load image2d<value::rgb8> input_1; @@ -165,128 +110,8 @@ int main(int argc, char *argv[]) input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_luma<value::int_u8>()); - // 1st simple binarization image2d<bool> - output = scribo::binarization::sauvola_ms(input_1_gl, w_1, s, k); - - mln::util::timer lt; - lt.start(); - - // Compute integral image - scribo::util::integral_sum_sum2_functor<value::int_u8,double> f_sum_sum2; - image2d<mln::util::couple<double,double> > - integral_sum_sum_2 = scribo::util::init_integral_image(input_1_gl, 1, f_sum_sum2); - - lt.stop(); - std::cout << "integral image - " << lt << std::endl; - lt.start(); - - // Find text lines - line_set<image2d<scribo::def::lbl_type> > - lines = scribo::text::extract_lines(output, c8()); - - typedef scribo::def::lbl_type V; - typedef image2d<V> L; - mln::util::array<unsigned> thickness(lines.nelements() + 1, 0); - const component_set<L>& comp_set = lines.components(); - const L& lbl = comp_set.labeled_image(); - - lt.stop(); - std::cout << "Text line finding - " << lt << std::endl; - lt.start(); - - // Compute run-lengths histogram in order to compute character - // thickness for each line. - for_all_lines(i, lines) - { - if (!lines(i).is_textline()) - continue; - - std::map<unsigned, unsigned> histo; - int count = 0; - for (int l = lines(i).bbox().pmin().row(); l <= lines(i).bbox().pmax().row(); ++l) - { - const V* end_ptr = &lbl.at_(l, lines(i).bbox().pmax().col() + 1); - for (const V* run_ptr = &lbl.at_(l, lines(i).bbox().pmin().col()); run_ptr != end_ptr; ++run_ptr) - if (*run_ptr) - ++count; - else - if (count) - { - if (histo.find(count) != histo.end()) - histo[count]++; - else - histo.insert(std::make_pair(count, 1)); - count = 0; - } - } - - unsigned max = 0; - unsigned thick = 0; - for (std::map<unsigned, unsigned>::const_iterator it = histo.begin(); it != histo.end(); ++it) - if (it->second > max) - { - max = it->second; - thick = it->first; - } - - thickness(lines(i).id()) = thick; - } - - lt.stop(); - std::cout << "run-lengths histogram - " << lt << std::endl; - - lt.start(); - // Compute thresholds for each pixel of each line and binarize again! - for_all_lines(i, lines) - { - if (!lines(i).is_textline()) - continue; - - double - win_min = thickness(lines(i).id()), - win_max = lines(i).bbox().height();; - - mln_assertion(win_min != 0); - mln_assertion(win_max != 0); - - scribo::binarization::internal::sauvola_formula formula; - - for (int row = lines(i).bbox().pmin().row(); - row <= lines(i).bbox().pmax().row(); - ++row) - { - bool* out_ptr = &output.at_(row, lines(i).bbox().pmin().col()); - value::int_u8* in_ptr = &input_1_gl.at_(row, lines(i).bbox().pmin().col()); - for (int col = lines(i).bbox().pmin().col(); - col <= lines(i).bbox().pmax().col(); - ++col) - { - // Min case - double T_min = compute_thres(integral_sum_sum_2, row, col, win_min, formula); - - // Max case - double T_max = compute_thres(integral_sum_sum_2, row, col, win_max, formula); - - // Final threshold - double teta = 0.3; // Good results from 0.1 to 0.3 according - // to the paper. - double T = teta * T_max + (1 - teta) * T_min; - - mln_assertion(T_min <= 255); - mln_assertion(T_max <= 255); - mln_assertion(T <= 255); - - *out_ptr++ = *in_ptr++ <= T; - } - } - } - - lt.stop(); - std::cout << "Last binarization - " << lt << std::endl; - - t.stop(); - std::cout << "Total time = " << t << std::endl; + output = scribo::binarization::kim(input_1_gl, w_1, k); - io::pbm::save(output, options.arg("output.pbm")); + io::magick::save(output, options.arg("output.*")); } diff --git a/scribo/tests/binarization/Makefile.am b/scribo/tests/binarization/Makefile.am index bac5a24..714bf93 100644 --- a/scribo/tests/binarization/Makefile.am +++ b/scribo/tests/binarization/Makefile.am @@ -21,6 +21,10 @@ include $(top_srcdir)/scribo/tests/tests.mk EXTRA_DIST = \ + kim.res.pbm \ + kim_wodd_heven.ref.pbm \ + kim_weven_hodd.ref.pbm \ + kim_wodd_hodd.ref.pbm \ niblack.res.pbm \ niblack_wodd_heven.ref.pbm \ niblack_weven_hodd.ref.pbm \ @@ -41,6 +45,7 @@ EXTRA_DIST = \ check_PROGRAMS = \ global_threshold \ + kim \ local_threshold \ niblack \ otsu \ @@ -50,6 +55,7 @@ check_PROGRAMS = \ global_threshold_SOURCES = global_threshold.cc +kim_SOURCES = kim.cc local_threshold_SOURCES = local_threshold.cc niblack_SOURCES = niblack.cc otsu_SOURCES = otsu.cc diff --git a/scribo/tests/binarization/sauvola.cc b/scribo/tests/binarization/kim.cc similarity index 77% copy from scribo/tests/binarization/sauvola.cc copy to scribo/tests/binarization/kim.cc index 54e8bbd..74e0eef 100644 --- a/scribo/tests/binarization/sauvola.cc +++ b/scribo/tests/binarization/kim.cc @@ -31,7 +31,7 @@ #include <mln/io/pgm/load.hh> #include <mln/io/pbm/load.hh> -#include <scribo/binarization/sauvola.hh> +#include <scribo/binarization/kim.hh> #include "tests/data.hh" @@ -44,10 +44,10 @@ int main() image2d<value::int_u8> input; io::pgm::load(input, MILENA_IMG_DIR "/lena.pgm"); - image2d<bool> bin = scribo::binarization::sauvola(input, 101); + image2d<bool> bin = scribo::binarization::kim(input, 101); image2d<bool> ref; - io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/sauvola.ref.pbm"); + io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/kim.ref.pbm"); mln_assertion(bin == ref); } @@ -57,10 +57,10 @@ int main() image2d<value::int_u8> input; io::pgm::load(input, SCRIBO_IMG_DIR "/lena_wodd_heven.pgm"); - image2d<bool> bin = scribo::binarization::sauvola(input, 101); + image2d<bool> bin = scribo::binarization::kim(input, 101); image2d<bool> ref; - io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/sauvola_wodd_heven.ref.pbm"); + io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/kim_wodd_heven.ref.pbm"); mln_assertion(bin == ref); } @@ -70,10 +70,10 @@ int main() image2d<value::int_u8> input; io::pgm::load(input, SCRIBO_IMG_DIR "/lena_weven_hodd.pgm"); - image2d<bool> bin = scribo::binarization::sauvola(input, 101); + image2d<bool> bin = scribo::binarization::kim(input, 101); image2d<bool> ref; - io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/sauvola_weven_hodd.ref.pbm"); + io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/kim_weven_hodd.ref.pbm"); mln_assertion(bin == ref); } @@ -83,10 +83,10 @@ int main() image2d<value::int_u8> input; io::pgm::load(input, SCRIBO_IMG_DIR "/lena_wodd_hodd.pgm"); - image2d<bool> bin = scribo::binarization::sauvola(input, 101); + image2d<bool> bin = scribo::binarization::kim(input, 101); image2d<bool> ref; - io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/sauvola_wodd_hodd.ref.pbm"); + io::pbm::load(ref, SCRIBO_TESTS_DIR "binarization/kim_wodd_hodd.ref.pbm"); mln_assertion(bin == ref); } diff --git a/scribo/tests/binarization/kim.ref.pbm b/scribo/tests/binarization/kim.ref.pbm new file mode 100644 index 0000000..d792130 Binary files /dev/null and b/scribo/tests/binarization/kim.ref.pbm differ diff --git a/scribo/tests/binarization/kim_weven_hodd.ref.pbm b/scribo/tests/binarization/kim_weven_hodd.ref.pbm new file mode 100644 index 0000000..45eab1d Binary files /dev/null and b/scribo/tests/binarization/kim_weven_hodd.ref.pbm differ diff --git a/scribo/tests/binarization/kim_wodd_heven.ref.pbm b/scribo/tests/binarization/kim_wodd_heven.ref.pbm new file mode 100644 index 0000000..1dc805f Binary files /dev/null and b/scribo/tests/binarization/kim_wodd_heven.ref.pbm differ diff --git a/scribo/tests/binarization/kim_wodd_hodd.ref.pbm b/scribo/tests/binarization/kim_wodd_hodd.ref.pbm new file mode 100644 index 0000000..5598763 Binary files /dev/null and b/scribo/tests/binarization/kim_wodd_hodd.ref.pbm differ -- 1.7.2.5