[Olena-patches] 3912: Small fixes in Scribo.

* preprocessing/all.hh, * all.hh: fix includes.

* binarization/simple.hh: fix documentation and warnings.

* make/text.hh: add routine signature.

* preprocessing/unskew.hh: remove useless debug.

* scribo.mk: update path to milena directory.

* tests/preprocessing/unskew.cc: new test for unskew routine.

* tests/unit_test/unit-tests.mk: update unit test list.

* text/clean.hh: enlarge the image before cleaning it up.

--- scribo/ChangeLog | 21 ++++++++++++++ scribo/all.hh | 12 ++++---- scribo/binarization/simple.hh | 24 +++++----------- scribo/make/text.hh | 19 +++++++++++-- scribo/preprocessing/all.hh | 4 +- scribo/preprocessing/unskew.hh | 1 - scribo/scribo.mk | 4 +- scribo/tests/preprocessing/unskew.cc | 17 ++++++++++++ scribo/tests/unit_test/unit-tests.mk | 38 +++++++++++++++----------- scribo/text/clean.hh | 49 +++++++++++++++++++++++++-------- 10 files changed, 131 insertions(+), 58 deletions(-) create mode 100644 scribo/tests/preprocessing/unskew.cc

diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 7b25394..4e49716 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,26 @@ 2009-05-28 Guillaume Lazzara lazzara@lrde.epita.fr

+ Small fixes in Scribo. + + * preprocessing/all.hh, + * all.hh: fix includes. + + * binarization/simple.hh: fix documentation and warnings. + + * make/text.hh: add routine signature. + + * preprocessing/unskew.hh: remove useless debug. + + * scribo.mk: update path to milena directory. + + * tests/preprocessing/unskew.cc: new test for unskew routine. + + * tests/unit_test/unit-tests.mk: update unit test list. + + * text/clean.hh: enlarge the image before cleaning it up. + +2009-05-28 Guillaume Lazzara lazzara@lrde.epita.fr + Update samples in Scribo.

* src/Makefile.am, diff --git a/scribo/all.hh b/scribo/all.hh index 80c4bbf..d78e6a2 100644 --- a/scribo/all.hh +++ b/scribo/all.hh @@ -1,4 +1,4 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory +// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -39,14 +39,14 @@ namespace scribo

} // end of namespace scribo

+# include <scribo/core/all.hh> +# include <scribo/debug/all.hh> # include <scribo/draw/all.hh> -# include <scribo/table/all.hh> -# include <scribo/util/all.hh> # include <scribo/filter/all.hh> # include <scribo/make/all.hh> +# include <scribo/preprocessing/all.hh> +# include <scribo/table/all.hh> # include <scribo/text/all.hh> -# include <scribo/core/all.hh> -# include <scribo/debug/all.hh> -# include <scribo/deskew/all.hh> +# include <scribo/util/all.hh>

#endif // ! SCRIBO_ALL_HH diff --git a/scribo/binarization/simple.hh b/scribo/binarization/simple.hh index 46953e9..36a5131 100644 --- a/scribo/binarization/simple.hh +++ b/scribo/binarization/simple.hh @@ -72,21 +72,13 @@ namespace scribo /// Simple binarization of a gray-level document. /*! * - * \param[in] input_ A binary image. - * \param[in] nbh_ The neighborhood used for labeling image components. - * \param[out] nlines The number of lines found. - * \param[in] vwin Window used to extract the vertical lines in a morphological - * opening - * \param[in] hwin Window used to extract the horizontal lines in a morphological - * opening + * \param[in] input_ A gray-level image. * - * \return pair of array of bounding boxes. The first array holds the - * vertical lines bounding boxes and the second one the - * horizontal lines bounding boxes. + * \result A Boolean image. */ template <typename I> mln_ch_value(I, bool) - simple(const Image<I>& input); + simple(const Image<I>& input_);

@@ -114,7 +106,7 @@ namespace scribo

unsigned nbasins; mln_ch_value(I, unsigned) w = morpho::watershed::flooding(g, c4(), nbasins); - h = histo::compute(input | (pw::value(w) == pw::cst(0))); + h = histo::compute(input | (pw::value(w) == pw::cst(0u))); }

@@ -128,13 +120,13 @@ namespace scribo convert::from_to(h, h_); hs_ = linear::gaussian_1d(h_, sigma, 0); l = labeling::regional_minima(hs_, c2(), n); - - if (n < 3) + + if (n < 3u) { std::cerr << "This method has not worked properly!" << std::endl;

debug::histo(h, "tmp_h.txt"); - + std::ofstream file("tmp_hs.txt"); mln_piter(box1d) p(h_.domain()); for_all(p) @@ -149,7 +141,7 @@ namespace scribo c[0] = point1d(0); // Force a neutral value for the non-label value (0). }

- int threshold; + unsigned threshold;

{ std::vector<int> v; diff --git a/scribo/make/text.hh b/scribo/make/text.hh index cac6356..406a77f 100644 --- a/scribo/make/text.hh +++ b/scribo/make/text.hh @@ -79,6 +79,20 @@ namespace scribo V& nbboxes);

+ /// Construct a new util::text from another one. + /// + /// \param[in] text A text structure. + /// \param[in] f_ A function v2v or v2b. It will be used to regroup text + /// components. + /// + /// \return a new util::text. + // + template <typename L, typename F> + scribo::util::text<L> + text(const scribo::util::text<L>& text, + const Function<F>& f_); + + # ifndef MLN_INCLUDE_ONLY

template <typename L> @@ -123,9 +137,7 @@ namespace scribo return make::text(cboxes, lbl, nbboxes); }

- //FIXME: we want the following routine to construct a new util::text - // from another one and a relabeling function. It avoids recomputing - // the whole underlying data (mass centers, bboxes...) + template <typename L, typename F> scribo::util::text<L> text(const scribo::util::text<L>& text, @@ -139,6 +151,7 @@ namespace scribo mln_value(L) new_nbboxes; mln::fun::i2v::array<mln_value(L)> fv2v = mln::make::relabelfun(f, text.nbboxes(), new_nbboxes); + L lbl = labeling::relabel(text.label_image(), text.nbboxes(), fv2v);

mln::util::array< accu::bbox<mln_site(L)> > tboxes(new_nbboxes.next()); diff --git a/scribo/preprocessing/all.hh b/scribo/preprocessing/all.hh index 145d6f4..0218eaa 100644 --- a/scribo/preprocessing/all.hh +++ b/scribo/preprocessing/all.hh @@ -1,4 +1,4 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory +// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) // // This file is part of the Olena Library. This library is free // software; you can redistribute it and/or modify it under the terms @@ -43,6 +43,6 @@ namespace scribo

} // end of namespace scribo

-# include <scribo/preprocessing/hough.hh> +# include <scribo/preprocessing/unskew.hh>

#endif // ! SCRIBO_PREPROCESSING_ALL_HH diff --git a/scribo/preprocessing/unskew.hh b/scribo/preprocessing/unskew.hh index bb71505..d092034 100644 --- a/scribo/preprocessing/unskew.hh +++ b/scribo/preprocessing/unskew.hh @@ -80,7 +80,6 @@ namespace scribo

point2d max_p = accu::compute(accu::max_site<image2d<float> >(), hough_ima);

- std::cout << max_p.col() << std::endl; double angle = 0; int max_angle = max_p.col();

diff --git a/scribo/scribo.mk b/scribo/scribo.mk index 866d51a..b32fe80 100644 --- a/scribo/scribo.mk +++ b/scribo/scribo.mk @@ -1,5 +1,5 @@ -include $(top_srcdir)/milena/tests/tests.mk +#include $(top_srcdir)/milena/tests/tests.mk

# Add path to Scribo's headers -AM_CPPFLAGS += -I$(top_srcdir)/ -I$(top_builddir)/ +AM_CPPFLAGS = -I$(top_srcdir)/ -I$(top_builddir)/ -I$(top_srcdir)/milena -I$(top_builddir)/milena -DNDEBUG -O2

diff --git a/scribo/tests/preprocessing/unskew.cc b/scribo/tests/preprocessing/unskew.cc new file mode 100644 index 0000000..c0291fe --- /dev/null +++ b/scribo/tests/preprocessing/unskew.cc @@ -0,0 +1,17 @@ +#include <mln/core/image/image2d.hh> +#include <mln/io/pbm/load.hh> +#include <mln/io/pbm/save.hh> + +#include <scribo/preprocessing/unskew.hh> + + +int main(int argc, char *argv[]) +{ + (void) argc; + using namespace mln; + + image2d<bool> ima; + io::pbm::load(ima, argv[1]); + + io::pbm::save(scribo::preprocessing::unskew(ima), "tr.pbm"); +} diff --git a/scribo/tests/unit_test/unit-tests.mk b/scribo/tests/unit_test/unit-tests.mk index f7204da..761253b 100644 --- a/scribo/tests/unit_test/unit-tests.mk +++ b/scribo/tests/unit_test/unit-tests.mk @@ -5,36 +5,42 @@ scribo_all \ scribo_binarization_simple \ scribo_core_all \ scribo_core_central_sites \ -scribo_core_component_bboxes \ scribo_core_erase_bboxes \ scribo_core_macros \ +scribo_core_object_image \ scribo_debug_all \ scribo_debug_save_label_image \ scribo_debug_save_linked_textbboxes_image \ scribo_debug_save_table_image \ scribo_debug_save_textbboxes_image \ +scribo_debug_usage \ scribo_demat \ +scribo_demat_old \ scribo_draw_all \ scribo_draw_bounding_box_links \ scribo_draw_bounding_boxes \ +scribo_extract_lines_discontinued \ scribo_extract_primitive_canvas \ scribo_extract_primitive_cells \ scribo_extract_primitive_lines_discontinued \ scribo_extract_primitive_lines_h_discontinued \ +scribo_extract_primitive_lines_h_thick \ scribo_extract_primitive_lines_thick \ scribo_extract_primitive_lines_v_discontinued \ +scribo_extract_primitive_lines_v_thick \ scribo_extract_primitive_objects \ scribo_filter_all \ -scribo_filter_large_components \ -scribo_filter_small_components \ -scribo_filter_thick_bboxes \ -scribo_filter_thin_bboxes \ +scribo_filter_large_objects \ +scribo_filter_small_objects \ +scribo_filter_thick_objects \ +scribo_filter_thin_objects \ scribo_make_all \ scribo_make_debug_filename \ scribo_make_influence_zone_graph \ scribo_make_text \ scribo_preprocessing_all \ scribo_preprocessing_unskew \ +scribo_src_woot \ scribo_table_align_lines_horizontaly \ scribo_table_align_lines_verticaly \ scribo_table_all \ @@ -42,8 +48,6 @@ scribo_table_connect_horizontal_lines \ scribo_table_connect_vertical_lines \ scribo_table_erase \ scribo_table_extract \ -scribo_table_extract_lines_with_opening \ -scribo_table_extract_lines_with_rank \ scribo_table_internal_align_lines \ scribo_table_internal_all \ scribo_table_internal_connect_lines \ @@ -53,7 +57,6 @@ scribo_table_repair_horizontal_lines \ scribo_table_repair_vertical_lines \ scribo_text_all \ scribo_text_clean \ -scribo_text_extract_bboxes \ scribo_text_extract_lines \ scribo_text_grouping_all \ scribo_text_grouping_group_from_double_link \ @@ -82,36 +85,42 @@ scribo_all_SOURCES = scribo_all.cc scribo_binarization_simple_SOURCES = scribo_binarization_simple.cc scribo_core_all_SOURCES = scribo_core_all.cc scribo_core_central_sites_SOURCES = scribo_core_central_sites.cc -scribo_core_component_bboxes_SOURCES = scribo_core_component_bboxes.cc scribo_core_erase_bboxes_SOURCES = scribo_core_erase_bboxes.cc scribo_core_macros_SOURCES = scribo_core_macros.cc +scribo_core_object_image_SOURCES = scribo_core_object_image.cc scribo_debug_all_SOURCES = scribo_debug_all.cc scribo_debug_save_label_image_SOURCES = scribo_debug_save_label_image.cc scribo_debug_save_linked_textbboxes_image_SOURCES = scribo_debug_save_linked_textbboxes_image.cc scribo_debug_save_table_image_SOURCES = scribo_debug_save_table_image.cc scribo_debug_save_textbboxes_image_SOURCES = scribo_debug_save_textbboxes_image.cc +scribo_debug_usage_SOURCES = scribo_debug_usage.cc scribo_demat_SOURCES = scribo_demat.cc +scribo_demat_old_SOURCES = scribo_demat_old.cc scribo_draw_all_SOURCES = scribo_draw_all.cc scribo_draw_bounding_box_links_SOURCES = scribo_draw_bounding_box_links.cc scribo_draw_bounding_boxes_SOURCES = scribo_draw_bounding_boxes.cc +scribo_extract_lines_discontinued_SOURCES = scribo_extract_lines_discontinued.cc scribo_extract_primitive_canvas_SOURCES = scribo_extract_primitive_canvas.cc scribo_extract_primitive_cells_SOURCES = scribo_extract_primitive_cells.cc scribo_extract_primitive_lines_discontinued_SOURCES = scribo_extract_primitive_lines_discontinued.cc scribo_extract_primitive_lines_h_discontinued_SOURCES = scribo_extract_primitive_lines_h_discontinued.cc +scribo_extract_primitive_lines_h_thick_SOURCES = scribo_extract_primitive_lines_h_thick.cc scribo_extract_primitive_lines_thick_SOURCES = scribo_extract_primitive_lines_thick.cc scribo_extract_primitive_lines_v_discontinued_SOURCES = scribo_extract_primitive_lines_v_discontinued.cc +scribo_extract_primitive_lines_v_thick_SOURCES = scribo_extract_primitive_lines_v_thick.cc scribo_extract_primitive_objects_SOURCES = scribo_extract_primitive_objects.cc scribo_filter_all_SOURCES = scribo_filter_all.cc -scribo_filter_large_components_SOURCES = scribo_filter_large_components.cc -scribo_filter_small_components_SOURCES = scribo_filter_small_components.cc -scribo_filter_thick_bboxes_SOURCES = scribo_filter_thick_bboxes.cc -scribo_filter_thin_bboxes_SOURCES = scribo_filter_thin_bboxes.cc +scribo_filter_large_objects_SOURCES = scribo_filter_large_objects.cc +scribo_filter_small_objects_SOURCES = scribo_filter_small_objects.cc +scribo_filter_thick_objects_SOURCES = scribo_filter_thick_objects.cc +scribo_filter_thin_objects_SOURCES = scribo_filter_thin_objects.cc scribo_make_all_SOURCES = scribo_make_all.cc scribo_make_debug_filename_SOURCES = scribo_make_debug_filename.cc scribo_make_influence_zone_graph_SOURCES = scribo_make_influence_zone_graph.cc scribo_make_text_SOURCES = scribo_make_text.cc scribo_preprocessing_all_SOURCES = scribo_preprocessing_all.cc scribo_preprocessing_unskew_SOURCES = scribo_preprocessing_unskew.cc +scribo_src_woot_SOURCES = scribo_src_woot.cc scribo_table_align_lines_horizontaly_SOURCES = scribo_table_align_lines_horizontaly.cc scribo_table_align_lines_verticaly_SOURCES = scribo_table_align_lines_verticaly.cc scribo_table_all_SOURCES = scribo_table_all.cc @@ -119,8 +128,6 @@ scribo_table_connect_horizontal_lines_SOURCES = scribo_table_connect_horizontal_ scribo_table_connect_vertical_lines_SOURCES = scribo_table_connect_vertical_lines.cc scribo_table_erase_SOURCES = scribo_table_erase.cc scribo_table_extract_SOURCES = scribo_table_extract.cc -scribo_table_extract_lines_with_opening_SOURCES = scribo_table_extract_lines_with_opening.cc -scribo_table_extract_lines_with_rank_SOURCES = scribo_table_extract_lines_with_rank.cc scribo_table_internal_align_lines_SOURCES = scribo_table_internal_align_lines.cc scribo_table_internal_all_SOURCES = scribo_table_internal_all.cc scribo_table_internal_connect_lines_SOURCES = scribo_table_internal_connect_lines.cc @@ -130,7 +137,6 @@ scribo_table_repair_horizontal_lines_SOURCES = scribo_table_repair_horizontal_li scribo_table_repair_vertical_lines_SOURCES = scribo_table_repair_vertical_lines.cc scribo_text_all_SOURCES = scribo_text_all.cc scribo_text_clean_SOURCES = scribo_text_clean.cc -scribo_text_extract_bboxes_SOURCES = scribo_text_extract_bboxes.cc scribo_text_extract_lines_SOURCES = scribo_text_extract_lines.cc scribo_text_grouping_all_SOURCES = scribo_text_grouping_all.cc scribo_text_grouping_group_from_double_link_SOURCES = scribo_text_grouping_group_from_double_link.cc diff --git a/scribo/text/clean.hh b/scribo/text/clean.hh index c8fa552..7a11329 100644 --- a/scribo/text/clean.hh +++ b/scribo/text/clean.hh @@ -34,14 +34,28 @@ /// Improve the quality of a text area.

# include <mln/core/concept/weighted_window.hh> +# include <mln/core/routine/extend.hh> + # include <mln/morpho/skeleton_constrained.hh> + # include <mln/topo/skeleton/is_simple_point.hh> # include <mln/topo/skeleton/crest.hh>

# include <mln/logical/not.hh>

-//# include <mln/debug/filename.hh> -//# include <mln/io/pbm/save.hh> +# include <mln/world/binary_2d/enlarge.hh> + +# include <mln/debug/filename.hh> +# include <mln/io/pbm/save.hh> + +#include <mln/labeling/wrap.hh> +#include <mln/win/octagon2d.hh> + +#include <mln/arith/revert.hh> + + +#include <mln/linear/gaussian.hh> +#include <mln/value/int_u8.hh>

namespace scribo { @@ -57,7 +71,7 @@ namespace scribo # ifndef MLN_INCLUDE_ONLY

-// static int plop = 0; + static int plop = 0;

template <typename I, typename W> mln_concrete(I) @@ -72,30 +86,41 @@ namespace scribo mln_precondition(input.is_valid()); mln_precondition(dmap_win.is_valid());

+ I input_large = world::binary_2d::enlarge(input, 2); + +// image2d<bool> blur = linear::gaussian(input_large, 2); +// image2dvalue::int_u8 blur = linear::gaussian(level::convert(value::int_u8(), input_large), 2); +// image2d<bool> blur = level::transform(linear::gaussian(level::convert(value::int_u8(), input_large), 2), fun::v2b::thresholdvalue::int_u8(100)); + mln_ch_value(I,unsigned) - dmap = transform::distance_front(logical::not_(input), c8(), + dmap = transform::distance_front(logical::not_(input_large), c8(), dmap_win, mln_max(unsigned)); +// io::pgm::save(labeling::wrap(dmap), mln::debug::filename("dmap.pgm"));

- I constraint = topo::skeleton::crest(input, dmap, c8()); +// I skeleton = topo::skeleton::crest(input_large, dmap, c8()); + I constraint = topo::skeleton::crest(input_large, dmap, c8()); mln_postcondition(constraint.is_valid());

+// io::pgm::save(labeling::wrap(constraint), mln::debug::filename("constraint.pgm")); + I skeleton = - morpho::skeleton_constrained(input, c8(), + morpho::skeleton_constrained(input_large, c8(), topo::skeleton::is_simple_point<I,neighb2d>, - extend(constraint, false), dmap); + extend(constraint, false), arith::revert(dmap));

- win::rectangle2d disk(3,3); + win::octagon2d disk(7); I output = morpho::dilation(skeleton, disk);

-// if (plop < 10) -// { +// if (plop > 20 && plop < 50) + { // io::pbm::save(input, mln::debug::filename("input.pbm")); +// io::pbm::save(input_large, mln::debug::filename("input_large_4x.pbm")); // io::pbm::save(skeleton, mln::debug::filename("skeleton.pbm")); // io::pbm::save(output, mln::debug::filename("dil_skel.pbm")); -// ++plop; -// } + }

+ ++plop; trace::exiting("scribo::text::clean"); return output; }