- Olena-patches

last-svn-commit-864-g1027cb2 scribo/text/paragraphs.hh: Remove unused argument.
by Guillaume Lazzara 06 May '11

06 May '11

--- scribo/ChangeLog | 4 ++++ scribo/scribo/text/paragraphs.hh | 5 ++--- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index b7e0f5a..2fb884b 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-05-06 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + * scribo/text/paragraphs.hh: Remove unused argument. + +2011-05-06 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Fix namespace ambiguities. * scribo/binarization/internal/first_pass_functor.hh, diff --git a/scribo/scribo/text/paragraphs.hh b/scribo/scribo/text/paragraphs.hh index 94e5751..6c9285b 100644 --- a/scribo/scribo/text/paragraphs.hh +++ b/scribo/scribo/text/paragraphs.hh @@ -64,8 +64,7 @@ namespace scribo void paragraph_links(const line_links<L>& left, const line_links<L>& right, line_links<L>& output, - const line_set<L>& lines, - const image2d<bool>& input) + const line_set<L>& lines) { output = left.duplicate(); @@ -1039,7 +1038,7 @@ namespace scribo // std::cout << "Finalizing merging" << std::endl; // finalize_line_merging(left, right, lines); std::cout << "Extracting paragraphs" << std::endl; - filter::paragraph_links(left, right, output, lines, input); + filter::paragraph_links(left, right, output, lines); paragraph_set<L> par_set = make::paragraph(output); return par_set; -- 1.5.6.5

1 0

last-svn-commit-863-g21452e4 Fix namespace ambiguities.
by Guillaume Lazzara 06 May '11

06 May '11

* scribo/binarization/internal/first_pass_functor.hh, * scribo/binarization/sauvola_ms.hh, * scribo/canvas/integral_browsing.hh, * scribo/convert/to_base64.hh, * scribo/preprocessing/denoise_bg.hh: Here. --- scribo/ChangeLog | 10 ++++++++ .../binarization/internal/first_pass_functor.hh | 6 ++-- scribo/scribo/binarization/sauvola_ms.hh | 24 ++++++++++---------- scribo/scribo/canvas/integral_browsing.hh | 10 ++++---- scribo/scribo/convert/to_base64.hh | 4 +- scribo/scribo/preprocessing/denoise_bg.hh | 5 ++- 6 files changed, 35 insertions(+), 24 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index d3a1780..b7e0f5a 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,15 @@ 2011-05-06 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Fix namespace ambiguities. + + * scribo/binarization/internal/first_pass_functor.hh, + * scribo/binarization/sauvola_ms.hh, + * scribo/canvas/integral_browsing.hh, + * scribo/convert/to_base64.hh, + * scribo/preprocessing/denoise_bg.hh: Here. + +2011-05-06 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Improve PDF output. * demo/xml2doc/image_crop.cc: Cleanup temporary files. diff --git a/scribo/scribo/binarization/internal/first_pass_functor.hh b/scribo/scribo/binarization/internal/first_pass_functor.hh index fb52be5..0b1a7ac 100644 --- a/scribo/scribo/binarization/internal/first_pass_functor.hh +++ b/scribo/scribo/binarization/internal/first_pass_functor.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. // @@ -64,7 +64,7 @@ namespace scribo image2d<value::int_u8> t_sub; unsigned n_nbhs; - util::array<int> dp; + mln::util::array<int> dp; double K_; diff --git a/scribo/scribo/binarization/sauvola_ms.hh b/scribo/scribo/binarization/sauvola_ms.hh index e8eb2c2..c1a3414 100644 --- a/scribo/scribo/binarization/sauvola_ms.hh +++ b/scribo/scribo/binarization/sauvola_ms.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. // @@ -138,7 +138,7 @@ namespace scribo unsigned lambda_min, unsigned lambda_max, unsigned s, unsigned q, unsigned i, unsigned w, - const image2d<util::couple<double,double> >& integral_sum_sum_2, + const image2d<mln::util::couple<double,double> >& integral_sum_sum_2, double K) { typedef image2d<int_u8> I; @@ -188,7 +188,7 @@ namespace scribo // 2nd pass { - util::array<mln_value_(I) *> ptr(ratio); + mln::util::array<mln_value_(I) *> ptr(ratio); unsigned nrows = geom::nrows(e_2); mln_box_runend_piter_(I) sp(sub.domain()); // Backward. @@ -272,7 +272,7 @@ namespace scribo template <typename I, typename J, typename K> mln_ch_value(I, bool) multi_scale_binarization(const I& in, const J& e2, - const util::array<K>& t_ima, + const mln::util::array<K>& t_ima, unsigned s) { mln_ch_value(I,bool) out; @@ -758,10 +758,10 @@ namespace scribo // Compute domains of subsampled images and make sure they can be // divided by 2. template <typename I> - util::array<util::couple<mln_domain(I), unsigned> > + mln::util::array<mln::util::couple<mln_domain(I), unsigned> > compute_sub_domains(const I& ima, unsigned n_scales, unsigned s) { - util::array<util::couple<unsigned, unsigned> > n(n_scales + 2); + mln::util::array<mln::util::couple<unsigned, unsigned> > n(n_scales + 2); n(1) = mln::make::couple(ima.nrows(), ima.ncols()); n(2) = mln::make::couple(sub(n(1).first(), s), @@ -771,7 +771,7 @@ namespace scribo sub(n(i - 1).second(), 2)); - util::array<util::couple<mln_domain(I), unsigned> > out(n.size()); + mln::util::array<mln::util::couple<mln_domain(I), unsigned> > out(n.size()); out(0) = mln::make::couple(mln::make::box2d(1,1), 1u); out(1) = mln::make::couple(mln::make::box2d(ima.nrows(), ima.ncols()), 2u); @@ -833,7 +833,7 @@ namespace scribo unsigned lambda_max_2 = lambda_min_2 * q; - util::array t_ima; + mln::util::array t_ima; // Make sure t_ima indexes start from 2. { @@ -842,7 +842,7 @@ namespace scribo t_ima.append(dummy); } - util::array sub_ima; + mln::util::array sub_ima; // Make sure sub_ima indexes start from 2. { @@ -851,7 +851,7 @@ namespace scribo sub_ima.append(dummy); } - util::array<util::couple<box2d, unsigned> > + mln::util::array<mln::util::couple<box2d, unsigned> > sub_domains = internal::compute_sub_domains(input_1, nb_subscale, s); border::adjust(input_1, sub_domains(1).second()); @@ -859,7 +859,7 @@ namespace scribo // Resize input and compute integral images. - typedef image2d<util::couple<double,double> > integral_t; + typedef image2d<mln::util::couple<double,double> > integral_t; integral_t integral_sum_sum_2; // Subsampling from scale 1 to 2. diff --git a/scribo/scribo/canvas/integral_browsing.hh b/scribo/scribo/canvas/integral_browsing.hh index 286dc53..0ec3d83 100644 --- a/scribo/scribo/canvas/integral_browsing.hh +++ b/scribo/scribo/canvas/integral_browsing.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. // @@ -40,7 +40,7 @@ namespace scribo template <typename F> - void integral_browsing(const image2d<util::couple<double, double> >& ima, + void integral_browsing(const image2d<mln::util::couple<double, double> >& ima, unsigned step, unsigned w, unsigned h, F& functor); @@ -75,13 +75,13 @@ namespace scribo template <typename F> - void integral_browsing(const image2d<util::couple<double, double> >& ima, + void integral_browsing(const image2d<mln::util::couple<double, double> >& ima, unsigned step, unsigned w, unsigned h, unsigned s, F& functor) { - typedef util::couple<double, double> V; + typedef mln::util::couple<double, double> V; typedef const V* Ptr; Ptr a_ima, b_ima, c_ima, d_ima; diff --git a/scribo/scribo/convert/to_base64.hh b/scribo/scribo/convert/to_base64.hh index 8df3ed2..81dda1f 100644 --- a/scribo/scribo/convert/to_base64.hh +++ b/scribo/scribo/convert/to_base64.hh @@ -81,7 +81,7 @@ namespace scribo template <typename I> void - to_base64(const Image& input, util::array<unsigned char>& output); + to_base64(const Image& input, mln::util::array<unsigned char>& output); # ifndef MLN_INCLUDE_ONLY @@ -116,7 +116,7 @@ namespace scribo template <typename I> void - to_base64(const Image& input_, util::array<unsigned char>& output) + to_base64(const Image& input_, mln::util::array<unsigned char>& output) { trace::entering("scribo::convert::to_base64"); diff --git a/scribo/scribo/preprocessing/denoise_bg.hh b/scribo/scribo/preprocessing/denoise_bg.hh index 8ad2d45..eda3803 100644 --- a/scribo/scribo/preprocessing/denoise_bg.hh +++ b/scribo/scribo/preprocessing/denoise_bg.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. // @@ -131,7 +132,7 @@ namespace scribo unsigned nlabels; image2d<unsigned> lbl = labeling::background(input, nbh, nlabels); - util::array<unsigned> + mln::util::array<unsigned> result = labeling::compute(accu::meta::math::count(), lbl, nlabels); mln::fun::i2v::array<bool> f(nlabels + 1, false); -- 1.5.6.5

1 0

last-svn-commit-862-ga1722e4 Improve PDF output.
by Guillaume Lazzara 06 May '11

06 May '11

* demo/xml2doc/image_crop.cc: Cleanup temporary files. * demo/xml2doc/templates/pdf/line.xsl: Use paragraph information. Take color into account. * demo/xml2doc/templates/pdf/regions.xsl: Fix image localization. --- scribo/ChangeLog | 11 + scribo/demo/xml2doc/image_crop.cc | 6 +- scribo/demo/xml2doc/templates/pdf/line.xsl | 390 +++++++++++++------------ scribo/demo/xml2doc/templates/pdf/regions.xsl | 14 +- 4 files changed, 215 insertions(+), 206 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index e30b230..d3a1780 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,14 @@ +2011-05-06 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + + Improve PDF output. + + * demo/xml2doc/image_crop.cc: Cleanup temporary files. + + * demo/xml2doc/templates/pdf/line.xsl: Use paragraph + information. Take color into account. + + * demo/xml2doc/templates/pdf/regions.xsl: Fix image localization. + 2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> Add hexadecimal color conversion routines. diff --git a/scribo/demo/xml2doc/image_crop.cc b/scribo/demo/xml2doc/image_crop.cc index 72fd2b4..4221ee7 100644 --- a/scribo/demo/xml2doc/image_crop.cc +++ b/scribo/demo/xml2doc/image_crop.cc @@ -40,11 +40,15 @@ ImageCrop::ImageCrop(const QString& xml, const QString& img, image_(img), output_dir_(output) { - mln::io::magick::load(ima_, image_.toStdString()); + if (!image_.isEmpty()) + mln::io::magick::load(ima_, image_.toStdString()); } ImageCrop::~ImageCrop() { + // Cleanup temporary files. + foreach (QString value, region_map_) + QFile::remove(value); } diff --git a/scribo/demo/xml2doc/templates/pdf/line.xsl b/scribo/demo/xml2doc/templates/pdf/line.xsl index 7056da0..4dda6c6 100644 --- a/scribo/demo/xml2doc/templates/pdf/line.xsl +++ b/scribo/demo/xml2doc/templates/pdf/line.xsl @@ -3,201 +3,205 @@  - - <xsl:template match="pcGts/page/text_region/line"> - -  - <xsl:variable name="x_height"> - <xsl:value-of select="@x_height" /> - </xsl:variable> - -  - <xsl:variable name="a_height"> - <xsl:value-of select="@a_height" /> - </xsl:variable> - - -  - <xsl:variable name="d_height_abs"> - <xsl:choose> - <xsl:when test="@d_height < 0"> - <xsl:value-of select="-@d_height" /> - </xsl:when> - <xsl:otherwise> - <xsl:value-of select="@d_height" /> - </xsl:otherwise> - </xsl:choose> - </xsl:variable> - -  + + <xsl:template match="pcGts/page/text_region"> + +  + <xsl:variable name="smallcase" select="'abcdefghijklmnopqrstuvwxyz'" /> + <xsl:variable name="uppercase" select="'ABCDEFGHIJKLMNOPQRSTUVWXYZ'" /> + +  + <xsl:variable name="colour"> + <xsl:choose> + <xsl:when test="@color != ''"> + <xsl:value-of select="translate(@color, $smallcase, $uppercase)" /> + </xsl:when> + <xsl:otherwise> + black + </xsl:otherwise> + </xsl:choose> + </xsl:variable> + + + <xsl:for-each select="line"> + +  + <xsl:variable name="x_height"> + <xsl:value-of select="@x_height" /> + </xsl:variable> + +  + <xsl:variable name="a_height"> + <xsl:value-of select="@a_height" /> + </xsl:variable> + + +  + <xsl:variable name="d_height_abs"> + <xsl:choose> + <xsl:when test="@d_height < 0"> + <xsl:value-of select="-@d_height" /> + </xsl:when> + <xsl:otherwise> + <xsl:value-of select="@d_height" /> + </xsl:otherwise> + </xsl:choose> + </xsl:variable> + +  - <xsl:variable name="a"> - <xsl:choose> - <xsl:when test="($a_height - $x_height) < $d_height_abs"> - <xsl:value-of select="$x_height + $d_height_abs" /> - </xsl:when> - <xsl:otherwise> - <xsl:value-of select="$a_height" /> - </xsl:otherwise> - </xsl:choose> - </xsl:variable> - -  - <xsl:variable name="d"> - <xsl:choose> - <xsl:when test="($a_height - $x_height) > $d_height_abs"> - <xsl:value-of select="$a_height - $x_height" /> - </xsl:when> - <xsl:otherwise> - <xsl:value-of select="$d_height_abs" /> - </xsl:otherwise> - </xsl:choose> - </xsl:variable> - -  - <xsl:variable name="y1"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@y" order="ascending" data-type="number"/> - <xsl:if test="position() = 1"> - <xsl:value-of select="@y" /> - </xsl:if> - </xsl:for-each> - </xsl:variable> - - <xsl:variable name="ymax"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@y" order="ascending" data-type="number"/> - <xsl:if test="position() = 3"> - <xsl:value-of select="@y" /> - </xsl:if> - </xsl:for-each> - </xsl:variable> - - <xsl:variable name="x1"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@x" order="ascending" data-type="number"/> - <xsl:if test="position() = 1"> - <xsl:value-of select="@x" /> - </xsl:if> - </xsl:for-each> - </xsl:variable> - - <xsl:variable name="xmax"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@x" order="ascending" data-type="number"/> - <xsl:if test="position() = 3"> - <xsl:value-of select="@x" /> - </xsl:if> - </xsl:for-each> - </xsl:variable> -  - -  - <xsl:variable name="smallcase" select="'abcdefghijklmnopqrstuvwxyz'" /> - <xsl:variable name="uppercase" select="'ABCDEFGHIJKLMNOPQRSTUVWXYZ'" /> - -  - <xsl:variable name="colour"> - <xsl:choose> - <xsl:when test="@txt_colour != ''"> - <xsl:value-of select="translate(@txt_colour, $smallcase, $uppercase)" /> - </xsl:when> - <xsl:otherwise> - black - </xsl:otherwise> - </xsl:choose> - </xsl:variable> - - - <fo:block-container position="absolute" border-width="5mm"> - - <xsl:attribute name="left"> - <xsl:value-of select="$x1" />px - </xsl:attribute> - <xsl:attribute name="top"> - <xsl:value-of select="$y1" />px - </xsl:attribute> - - <xsl:attribute name="right"> - <xsl:value-of select="$xmax" />px - </xsl:attribute> - <xsl:attribute name="bottom"> - <xsl:value-of select="$ymax" />px - </xsl:attribute> - - - <xsl:attribute name="width"> - <xsl:value-of select="$xmax - $x1" />px - </xsl:attribute> - - <xsl:attribute name="color"> - <xsl:value-of select="$colour" /> - </xsl:attribute> - -  - -  - -  - <fo:table table-layout="fixed"> - - <xsl:attribute name="width"> - <xsl:value-of select="$xmax -$x1" />px - </xsl:attribute> - - <fo:table-column column-number="1"> - <xsl:attribute name="column-width"> - <xsl:value-of select="$xmax -$x1" />px - </xsl:attribute> - </fo:table-column> - - <fo:table-body start-indent="0pt" text-align="justify" text-align-last="justify"> - - <fo:table-row> - - <fo:table-cell> - - <fo:block font-family="Times" wrap-option="no-wrap" white-space-collapse="true" text-align-last="justify" text-align="justify"> - - <xsl:attribute name="font-size"> - <xsl:choose> - <xsl:when test="($a + $d) > 50"> - <xsl:value-of select="0.82 * ($a + $d)" />px - </xsl:when> - <xsl:otherwise> - <xsl:value-of select="0.95 * ($a + $d)" />px - </xsl:otherwise> - </xsl:choose> - </xsl:attribute> - - <xsl:value-of select="@text"/> - </fo:block> - - </fo:table-cell> - - </fo:table-row> - - </fo:table-body> - - </fo:table> - - - </fo:block-container> + <xsl:variable name="d"> + <xsl:choose> + <xsl:when test="($a_height - $x_height) > $d_height_abs"> + <xsl:value-of select="$a_height - $x_height" /> + </xsl:when> + <xsl:otherwise> + <xsl:value-of select="$d_height_abs" /> + </xsl:otherwise> + </xsl:choose> + </xsl:variable> + +  + <xsl:variable name="y1"> + <xsl:for-each select="coords/point"> + <xsl:sort select="@y" order="ascending" data-type="number"/> + <xsl:if test="position() = 1"> + <xsl:value-of select="@y" /> + </xsl:if> + </xsl:for-each> + </xsl:variable> + + <xsl:variable name="ymax"> + <xsl:for-each select="coords/point"> + <xsl:sort select="@y" order="ascending" data-type="number"/> + <xsl:if test="position() = 3"> + <xsl:value-of select="@y" /> + </xsl:if> + </xsl:for-each> + </xsl:variable> + + <xsl:variable name="x1"> + <xsl:for-each select="coords/point"> + <xsl:sort select="@x" order="ascending" data-type="number"/> + <xsl:if test="position() = 1"> + <xsl:value-of select="@x" /> + </xsl:if> + </xsl:for-each> + </xsl:variable> + + <xsl:variable name="xmax"> + <xsl:for-each select="coords/point"> + <xsl:sort select="@x" order="ascending" data-type="number"/> + <xsl:if test="position() = 3"> + <xsl:value-of select="@x" /> + </xsl:if> + </xsl:for-each> + </xsl:variable> +  + + <fo:block-container position="absolute" border-width="5mm"> + + <xsl:attribute name="left"> + <xsl:value-of select="$x1" />px + </xsl:attribute> + <xsl:attribute name="top"> + <xsl:value-of select="$y1" />px + </xsl:attribute> + + <xsl:attribute name="right"> + <xsl:value-of select="$xmax" />px + </xsl:attribute> + <xsl:attribute name="bottom"> + <xsl:value-of select="$ymax" />px + </xsl:attribute> + + + <xsl:attribute name="width"> + <xsl:value-of select="$xmax - $x1" />px + </xsl:attribute> + + <xsl:attribute name="color"> + <xsl:value-of select="$colour" /> + </xsl:attribute> + +  + +  + +  + <fo:table table-layout="fixed"> + + <xsl:attribute name="width"> + <xsl:value-of select="$xmax -$x1" />px + </xsl:attribute> + + <fo:table-column column-number="1"> + <xsl:attribute name="column-width"> + <xsl:value-of select="$xmax -$x1" />px + </xsl:attribute> + </fo:table-column> + + <fo:table-body start-indent="0pt" text-align="justify" text-align-last="justify"> + + <fo:table-row> + + <fo:table-cell> + + <fo:block font-family="Times" wrap-option="no-wrap" white-space-collapse="true" text-align-last="justify" text-align="justify"> + + <xsl:attribute name="font-size"> + <xsl:choose> + <xsl:when test="($a + $d) > 50"> + <xsl:value-of select="0.82 * ($a + $d)" />px + </xsl:when> + <xsl:otherwise> + <xsl:value-of select="0.95 * ($a + $d)" />px + </xsl:otherwise> + </xsl:choose> + </xsl:attribute> + + <xsl:value-of select="@text"/> + </fo:block> + + </fo:table-cell> + + </fo:table-row> + + </fo:table-body> + + </fo:table> + + + </fo:block-container> + + </xsl:for-each> </xsl:template> diff --git a/scribo/demo/xml2doc/templates/pdf/regions.xsl b/scribo/demo/xml2doc/templates/pdf/regions.xsl index f236a1f..7772cbf 100644 --- a/scribo/demo/xml2doc/templates/pdf/regions.xsl +++ b/scribo/demo/xml2doc/templates/pdf/regions.xsl @@ -18,21 +18,11 @@  <xsl:variable name="y1"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@y" order="ascending" data-type="number"/> - <xsl:if test="position() = 1"> - <xsl:value-of select="@y" /> - </xsl:if> - </xsl:for-each> + <xsl:value-of select="@y_min" /> </xsl:variable> <xsl:variable name="x1"> - <xsl:for-each select="coords/point"> - <xsl:sort select="@x" order="ascending" data-type="number"/> - <xsl:if test="position() = 1"> - <xsl:value-of select="@x" /> - </xsl:if> - </xsl:for-each> + <xsl:value-of select="@x_min" /> </xsl:variable>  -- 1.5.6.5

1 0

last-svn-commit-861-g41f3675 configure.ac: Configure scribo/tests/util.
by Guillaume Lazzara 05 May '11

05 May '11

--- ChangeLog | 4 ++++ configure.ac | 3 +++ 2 files changed, 7 insertions(+), 0 deletions(-) diff --git a/ChangeLog b/ChangeLog index 31d498c..852ea6e 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,7 @@ +2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + + * configure.ac: Configure scribo/tests/util. + 2011-04-05 Guillaume Lazzara <z(a)lrde.epita.fr> * configure.ac: Configure scribo/tests/estim. diff --git a/configure.ac b/configure.ac index 189d905..fdbddd9 100644 --- a/configure.ac +++ b/configure.ac @@ -348,6 +348,8 @@ AC_CONFIG_FILES([ scribo/tests/core/Makefile scribo/tests/estim/Makefile scribo/tests/filter/Makefile + scribo/tests/io/Makefile + scribo/tests/io/xml/Makefile scribo/tests/preprocessing/Makefile scribo/tests/primitive/Makefile scribo/tests/primitive/extract/Makefile @@ -357,6 +359,7 @@ AC_CONFIG_FILES([ scribo/tests/toolchain/Makefile scribo/tests/toolchain/nepomuk/Makefile scribo/tests/unit_test/Makefile + scribo/tests/util/Makefile ]) # Flags for Scribo -- 1.5.6.5

1 0

last-svn-commit-859-gd90f512 Explicitly set vertical or horizontal attribute for separators in XML output.
by Guillaume Lazzara 05 May '11

05 May '11

* demo/xml2doc/image_crop.cc, * demo/xml2doc/templates/pdf/regions.xsl: Handle new XML tags. * scribo/core/document.hh, * scribo/core/tag/component.hh, * scribo/io/img/internal/debug_img_visitor.hh, * scribo/io/img/internal/full_img_visitor.hh, * scribo/io/img/internal/non_text_img_visitor.hh, * scribo/io/xml/internal/extended_page_xml_visitor.hh, * scribo/io/xml/internal/full_xml_visitor.hh, * scribo/io/xml/internal/page_xml_visitor.hh: Update enum declaration and handle new cases. --- scribo/ChangeLog | 18 ++++++++ scribo/demo/xml2doc/image_crop.cc | 7 ++- scribo/demo/xml2doc/templates/pdf/regions.xsl | 3 +- scribo/scribo/core/document.hh | 4 +- scribo/scribo/core/tag/component.hh | 16 +++++-- scribo/scribo/io/img/internal/debug_img_visitor.hh | 3 +- scribo/scribo/io/img/internal/full_img_visitor.hh | 3 +- .../scribo/io/img/internal/non_text_img_visitor.hh | 3 +- .../io/xml/internal/extended_page_xml_visitor.hh | 24 +++++++++- scribo/scribo/io/xml/internal/full_xml_visitor.hh | 45 +++++++++++++++++--- scribo/scribo/io/xml/internal/page_xml_visitor.hh | 3 +- 11 files changed, 105 insertions(+), 24 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index efc42cf..e857752 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,23 @@ 2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Explicitly set vertical or horizontal attribute for separators in + XML output. + + * demo/xml2doc/image_crop.cc, + * demo/xml2doc/templates/pdf/regions.xsl: Handle new XML tags. + + * scribo/core/document.hh, + * scribo/core/tag/component.hh, + * scribo/io/img/internal/debug_img_visitor.hh, + * scribo/io/img/internal/full_img_visitor.hh, + * scribo/io/img/internal/non_text_img_visitor.hh, + * scribo/io/xml/internal/extended_page_xml_visitor.hh, + * scribo/io/xml/internal/full_xml_visitor.hh, + * scribo/io/xml/internal/page_xml_visitor.hh: Update enum + declaration and handle new cases. + +2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Make use of opacity in PDF output. * demo/xml2doc/image_crop.cc: Create opacity mask. diff --git a/scribo/demo/xml2doc/image_crop.cc b/scribo/demo/xml2doc/image_crop.cc index 7d61f75..72fd2b4 100644 --- a/scribo/demo/xml2doc/image_crop.cc +++ b/scribo/demo/xml2doc/image_crop.cc @@ -101,7 +101,7 @@ void ImageCrop::from_base64() child = child.firstChild(); while (!child.isNull()) { - if (child.toElement().tagName().contains(QRegExp("(image|graphic|chart|separator|table)_region"))) + if (child.toElement().tagName().contains(QRegExp("(image|graphic|chart|horizontal_separator|vertical_separator|table)_region"))) { QDomNode node = child.firstChild(); QString id = child.toElement().attribute("id", "none"); @@ -208,7 +208,7 @@ void ImageCrop::to_base64(const QString& out_file, bool no_crop) while(!line.contains("</pcGts>")) { stream2 << "\n" << line; - if (line.contains(QRegExp("<(image|graphic|chart|separator|table)_region"))) + if (line.contains(QRegExp("<(image|graphic|chart|horizontal_separator|vertical_separator|table)_region"))) { stream2 << "\n" << " <container>\n"; stream2 << " <mime>png</mime>\n"; @@ -268,7 +268,7 @@ bool ImageCrop::crop_regions(bool temp) while (!region.isNull()) { - if (region.toElement().tagName().contains(QRegExp("(image|graphic|chart|separator|table)_region"))) + if (region.toElement().tagName().contains(QRegExp("(image|graphic|chart|vertical_separator|horizontal_separator|table)_region"))) { found_regions = true; @@ -338,6 +338,7 @@ bool ImageCrop::crop_regions(bool temp) else io::magick::save(crop, opacity_mask, QString(output_dir_ + id + ".png").toStdString()); } + region = region.nextSibling(); } diff --git a/scribo/demo/xml2doc/templates/pdf/regions.xsl b/scribo/demo/xml2doc/templates/pdf/regions.xsl index add0cba..f236a1f 100644 --- a/scribo/demo/xml2doc/templates/pdf/regions.xsl +++ b/scribo/demo/xml2doc/templates/pdf/regions.xsl @@ -8,7 +8,8 @@ pcGts/page/graphic_region| pcGts/page/chart_region| pcGts/page/table_region| - pcGts/page/separator_region"> + pcGts/page/vertical_separator_region| + pcGts/page/horizontal_separator_region">  <xsl:variable name="data"> diff --git a/scribo/scribo/core/document.hh b/scribo/scribo/core/document.hh index e287c1d..0fe2be3 100644 --- a/scribo/scribo/core/document.hh +++ b/scribo/scribo/core/document.hh @@ -343,7 +343,7 @@ namespace scribo mln_value(L) ncomps; hline_seps_comps_ = primitive::extract::components(hline_seps, mln::c8(), ncomps, - component::LineSeparator); + component::HorizontalLineSeparator); } @@ -390,7 +390,7 @@ namespace scribo mln_value(L) ncomps; vline_seps_comps_ = primitive::extract::components(vline_seps, mln::c8(), ncomps, - component::LineSeparator); + component::VerticalLineSeparator); } diff --git a/scribo/scribo/core/tag/component.hh b/scribo/scribo/core/tag/component.hh index 7cd2ede..dc9db90 100644 --- a/scribo/scribo/core/tag/component.hh +++ b/scribo/scribo/core/tag/component.hh @@ -55,7 +55,8 @@ namespace scribo { Undefined = 0, Character, - LineSeparator, + VerticalLineSeparator, + HorizontalLineSeparator, WhitespaceSeparator, Noise, Punctuation, @@ -117,8 +118,11 @@ namespace scribo case Character: str = "Character"; break; - case LineSeparator: - str = "LineSeparator"; + case HorizontalLineSeparator: + str = "HorizontalLineSeparator"; + break; + case VerticalLineSeparator: + str = "VerticalLineSeparator"; break; case WhitespaceSeparator: str = "WhitespaceSeparator"; @@ -143,8 +147,10 @@ namespace scribo { if (str == "Character") return Character; - else if (str == "LineSeparator") - return LineSeparator; + else if (str == "HorizontalLineSeparator") + return HorizontalLineSeparator; + else if (str == "VerticalLineSeparator") + return VerticalLineSeparator; else if (str == "WhitespaceSeparator") return WhitespaceSeparator; else if (str == "Noise") diff --git a/scribo/scribo/io/img/internal/debug_img_visitor.hh b/scribo/scribo/io/img/internal/debug_img_visitor.hh index f1c689d..62097f5 100644 --- a/scribo/scribo/io/img/internal/debug_img_visitor.hh +++ b/scribo/scribo/io/img/internal/debug_img_visitor.hh @@ -179,7 +179,8 @@ namespace scribo { switch (info.type()) { - case component::LineSeparator: + case component::HorizontalLineSeparator: + case component::VerticalLineSeparator: { mln::draw::box(output, compute_bbox(info.bbox()), literal::cyan); diff --git a/scribo/scribo/io/img/internal/full_img_visitor.hh b/scribo/scribo/io/img/internal/full_img_visitor.hh index 30987db..170b6a1 100644 --- a/scribo/scribo/io/img/internal/full_img_visitor.hh +++ b/scribo/scribo/io/img/internal/full_img_visitor.hh @@ -135,7 +135,8 @@ namespace scribo { switch (info.type()) { - case component::LineSeparator: + case component::HorizontalLineSeparator: + case component::VerticalLineSeparator: { mln::draw::box(output, info.bbox(), literal::cyan); } diff --git a/scribo/scribo/io/img/internal/non_text_img_visitor.hh b/scribo/scribo/io/img/internal/non_text_img_visitor.hh index 24b027e..cc1acb9 100644 --- a/scribo/scribo/io/img/internal/non_text_img_visitor.hh +++ b/scribo/scribo/io/img/internal/non_text_img_visitor.hh @@ -132,7 +132,8 @@ namespace scribo { switch (info.type()) { - case component::LineSeparator: + case component::HorizontalLineSeparator: + case component::VerticalLineSeparator: { mln::draw::box(output, info.bbox(), literal::cyan); } diff --git a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh index eaf08ac..83e5b0b 100644 --- a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh @@ -179,9 +179,9 @@ namespace scribo break; } - case component::LineSeparator: + case component::VerticalLineSeparator: { - output << " <separator_region id=\"sr" << info.id() + output << " <vertical_separator_region id=\"vlsr" << info.id() << "\" sep_orientation=\"0.000000\" " << " sep_colour=\"Black\" " << " sep_bgcolour=\"White\"" @@ -193,7 +193,25 @@ namespace scribo internal::print_box_coords(output, info.bbox(), " "); - output << " </separator_region>" << std::endl; + output << " </vertical_separator_region>" << std::endl; + break; + } + + case component::HorizontalLineSeparator: + { + output << " <horizontal_separator_region id=\"hlsr" << info.id() + << "\" sep_orientation=\"0.000000\" " + << " sep_colour=\"Black\" " + << " sep_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </horizontal_separator_region>" << std::endl; break; } diff --git a/scribo/scribo/io/xml/internal/full_xml_visitor.hh b/scribo/scribo/io/xml/internal/full_xml_visitor.hh index c294bbc..b920ca8 100644 --- a/scribo/scribo/io/xml/internal/full_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/full_xml_visitor.hh @@ -355,7 +355,12 @@ namespace scribo { output << " <whitespace_separator_region id=\"wss" << info.id() - << "\">" << std::endl; + << "\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_box_coords(output, info.bbox(), " "); @@ -363,16 +368,39 @@ namespace scribo break; } - case component::LineSeparator: + case component::VerticalLineSeparator: { - output << " <separator_region id=\"sr" << info.id() + output << " <vertical_separator_region id=\"vlsr" << info.id() << "\" sep_orientation=\"0.000000\" " << " sep_colour=\"Black\" " - << " sep_bgcolour=\"White\">" << std::endl; + << " sep_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_box_coords(output, info.bbox(), " "); - output << " </separator_region>" << std::endl; + output << " </vertical_separator_region>" << std::endl; + break; + } + + case component::HorizontalLineSeparator: + { + output << " <horizontal_separator_region id=\"hlsr" << info.id() + << "\" sep_orientation=\"0.000000\" " + << " sep_colour=\"Black\" " + << " sep_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </horizontal_separator_region>" << std::endl; break; } @@ -383,7 +411,12 @@ namespace scribo << "\" img_colour_type=\"24_Bit_Colour\"" << " img_orientation=\"0.000000\" " << " img_emb_text=\"No\" " - << " img_bgcolour=\"White\">" << std::endl; + << " img_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_box_coords(output, info.bbox(), " "); diff --git a/scribo/scribo/io/xml/internal/page_xml_visitor.hh b/scribo/scribo/io/xml/internal/page_xml_visitor.hh index 19665d6..0014caf 100644 --- a/scribo/scribo/io/xml/internal/page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/page_xml_visitor.hh @@ -146,7 +146,8 @@ namespace scribo { switch (info.type()) { - case component::LineSeparator: + case component::VerticalLineSeparator: + case component::HorizontalLineSeparator: { output << " <separator_region id=\"sr" << info.id() << "\" sep_orientation=\"0.000000\" " -- 1.5.6.5

1 0

last-svn-commit-858-gbc3f2b6 Make use of opacity in PDF output.
by Guillaume Lazzara 05 May '11

05 May '11

* demo/xml2doc/image_crop.cc: Create opacity mask. * scribo/io/xml/internal/extended_page_xml_visitor.hh: Add bbox information in region tag. --- scribo/ChangeLog | 9 +++ scribo/demo/xml2doc/image_crop.cc | 60 +++++++++++++------- .../io/xml/internal/extended_page_xml_visitor.hh | 21 ++++++- 3 files changed, 67 insertions(+), 23 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index a213078..efc42cf 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,12 @@ +2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + + Make use of opacity in PDF output. + + * demo/xml2doc/image_crop.cc: Create opacity mask. + + * scribo/io/xml/internal/extended_page_xml_visitor.hh: Add bbox + information in region tag. + 2011-05-04 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> Small fixes in Scribo. diff --git a/scribo/demo/xml2doc/image_crop.cc b/scribo/demo/xml2doc/image_crop.cc index ba30d8a..7d61f75 100644 --- a/scribo/demo/xml2doc/image_crop.cc +++ b/scribo/demo/xml2doc/image_crop.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. // @@ -16,10 +17,13 @@ #include <limits.h> +#include <scribo/core/macros.hh> #include <scribo/preprocessing/crop.hh> #include <mln/value/rgb8.hh> #include <mln/core/alias/box2d.hh> #include <mln/core/image/image2d.hh> +#include <mln/util/couple.hh> +#include <mln/draw/line.hh> #include <mln/io/magick/save.hh> #include <mln/io/magick/load.hh> #include <mln/io/ppm/all.hh> @@ -28,6 +32,7 @@ #include "loader.hh" #include "common.hh" +#include <mln/io/pbm/save.hh> ImageCrop::ImageCrop(const QString& xml, const QString& img, const QString& output) @@ -270,7 +275,14 @@ bool ImageCrop::crop_regions(bool temp) QDomNode coords = region.firstChild(); QString id = region.toElement().attribute("id", "none"); - // qDebug() << region.toElement().tagName(); + // Retrieve region bbox. + using namespace mln; + def::coord + x_min = region.toElement().attribute("x_min").toInt(), + y_min = region.toElement().attribute("y_min").toInt(), + x_max = region.toElement().attribute("x_max").toInt(), + y_max = region.toElement().attribute("y_max").toInt(); + box2d box = make::box2d(y_min, x_min, y_max, x_max); while (!coords.isNull() && !coords.toElement().tagName().contains("coords")) coords = coords.nextSibling(); @@ -279,44 +291,52 @@ bool ImageCrop::crop_regions(bool temp) break; QDomNode point = coords.firstChild(); - int x_max = 0; - int y_max = 0; - int x_min = INT_MAX; - int y_min = INT_MAX; + // For each row, store first and last column of the line + // to be set as object. The rest will be considered as + // transparent. + util::array<util::couple<def::coord, def::coord> > + p_mask(box.nrows(), + util::couple<def::coord, def::coord>(x_max + 1, x_min - 1)); + + // Compute opacity mask for image region. while (!point.isNull()) { int x = point.toElement().attribute("x", "none").toInt(); int y = point.toElement().attribute("y", "none").toInt(); - if (x < x_min) - x_min = x; - if (x > x_max) - x_max = x; - - if (y < y_min) - y_min = y; - if (y > y_max) - y_max = y; + if (p_mask(y - y_min).first() > x) + p_mask(y - y_min).first() = x; + if (p_mask(y - y_min).second() < x) + p_mask(y - y_min).second() = x; point = point.nextSibling(); } - using namespace mln; - box2d box = make::box2d(y_min, x_min, y_max, x_max); - + // Crop image in input. image2d<value::rgb8> crop = scribo::preprocessing::crop(ima_, box); + // Build image mask. + image2d<bool> opacity_mask(box); + { + data::fill(opacity_mask, true); + for_all_elements(e, p_mask) + draw::line(opacity_mask, + point2d(e + y_min, p_mask(e).first()), + point2d(e + y_min, p_mask(e).second()), + false); + } + if (temp) { QTemporaryFile tmp(output_dir_ + id + ".XXXXXX.png"); tmp.open(); region_map_[id] = tmp.fileName(); tmp.setAutoRemove(false); - io::magick::save(crop, tmp.fileName().toStdString()); + io::magick::save(crop, opacity_mask, tmp.fileName().toStdString()); } else - io::magick::save(crop, QString(output_dir_ + id + ".png").toStdString()); + io::magick::save(crop, opacity_mask, QString(output_dir_ + id + ".png").toStdString()); } region = region.nextSibling(); } diff --git a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh index b463677..eaf08ac 100644 --- a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh @@ -166,7 +166,12 @@ namespace scribo { output << " <whitespace_separator_region id=\"wss" << info.id() - << "\">" << std::endl; + << "\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_box_coords(output, info.bbox(), " "); @@ -179,7 +184,12 @@ namespace scribo output << " <separator_region id=\"sr" << info.id() << "\" sep_orientation=\"0.000000\" " << " sep_colour=\"Black\" " - << " sep_bgcolour=\"White\">" << std::endl; + << " sep_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_box_coords(output, info.bbox(), " "); @@ -195,7 +205,12 @@ namespace scribo << "\" img_colour_type=\"24_Bit_Colour\"" << " img_orientation=\"0.000000\" " << " img_emb_text=\"No\" " - << " img_bgcolour=\"White\">" << std::endl; + << " img_bgcolour=\"White\"" + << " x_min=\"" << info.bbox().pmin().col() << "\"" + << " y_min=\"" << info.bbox().pmin().row() << "\"" + << " x_max=\"" << info.bbox().pmax().col() << "\"" + << " y_max=\"" << info.bbox().pmax().row() << "\"" + << ">" << std::endl; internal::print_image_coords(output, ((elt_edge | info.bbox()) -- 1.5.6.5

1 0

last-svn-commit-857-g9b5c9ab mln/io/magick/save.hh: Add support for opacity.
by Guillaume Lazzara 05 May '11

05 May '11

--- milena/ChangeLog | 4 + milena/mln/io/magick/save.hh | 155 ++++++++++++++++++++++++++++++++++-------- 2 files changed, 131 insertions(+), 28 deletions(-) diff --git a/milena/ChangeLog b/milena/ChangeLog index 8008503..d12cc20 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,3 +1,7 @@ +2011-05-05 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + + * mln/io/magick/save.hh: Add support for opacity. + 2011-04-05 Guillaume Lazzara <z(a)lrde.epita.fr> * mln/accu/stat/median_h.hh: Add missing operator=(). diff --git a/milena/mln/io/magick/save.hh b/milena/mln/io/magick/save.hh index 2ea2957..f592059 100644 --- a/milena/mln/io/magick/save.hh +++ b/milena/mln/io/magick/save.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. // @@ -59,13 +59,29 @@ namespace mln { /** Save a Milena image into a file using Magick++. + \overload - \param[out] ima The image to save. - \param[in] filename The name of the output file. */ + \param[in] ima The image to save. + \param[in] filename The name of the output file. */ template <typename I> void save(const Image& ima, const std::string& filename); + /** Save a Milena image into a file using Magick++. + + \param[in] ima The image to save. + + \param[in] opacity_mask Mask used to set pixel opacity_mask in output + image. Output format must support this feature to be taken + into account. + + \param[in] filename The name of the output file. + */ + template <typename I, typename J> + void + save(const Image& ima, const Image<J>& opacity_mask, + const std::string& filename); + // FIXME: Unfinished? #if 0 @@ -125,10 +141,92 @@ namespace mln } // end of namespace mln::io::magick::impl - template <typename I> - inline + namespace internal + { + + template <typename I> + void + paste_data(const Image& ima_, Magick::Image& magick_ima) + { + const I& ima = exact(ima_); + + def::coord + minrow = geom::min_row(ima), + mincol = geom::min_col(ima), + maxrow = geom::max_row(ima), + maxcol = geom::max_col(ima), + ncols = geom::ncols(ima), + nrows = geom::nrows(ima); + + // Ensure that there is only one reference to underlying image + // If this is not done, then image pixels will not be modified. + magick_ima.modifyImage(); + + Magick::Pixels view(magick_ima); + // As above, `ncols' is passed before `nrows'. + Magick::PixelPacket* pixels = view.get(0, 0, ncols, nrows); + const mln_value(I) *ptr_ima = &ima(ima.domain().pmin()); + + unsigned row_offset = ima.delta_index(dpoint2d(+1, - ncols)); + + for (def::coord row = minrow; row <= maxrow; + ++row, ptr_ima += row_offset) + for (def::coord col = mincol; col <= maxcol; ++col) + *pixels++ = impl::get_color(*ptr_ima++); + + view.sync(); + } + + template <typename I, typename J> + void + paste_data_opacity(const Image& ima_, + const Image<J>& opacity_mask_, + Magick::Image& magick_ima) + { + const I& ima = exact(ima_); + const J& opacity_mask = exact(opacity_mask_); + + def::coord + minrow = geom::min_row(ima), + mincol = geom::min_col(ima), + maxrow = geom::max_row(ima), + maxcol = geom::max_col(ima), + ncols = geom::ncols(ima), + nrows = geom::nrows(ima); + + // Ensure that there is only one reference to underlying image + // If this is not done, then image pixels will not be modified. + magick_ima.modifyImage(); + + Magick::Pixels view(magick_ima); + // As above, `ncols' is passed before `nrows'. + Magick::PixelPacket* pixels = view.get(0, 0, ncols, nrows); + const mln_value(I) *ptr_ima = &ima(ima.domain().pmin()); + const mln_value(J) *ptr_opacity_mask = &opacity_mask(opacity_mask.domain().pmin()); + + unsigned row_offset = ima.delta_index(dpoint2d(+1, - ncols)); + unsigned opacity_row_offset = opacity_mask.delta_index(dpoint2d(+1, - ncols)); + + for (def::coord row = minrow; row <= maxrow; + ++row, ptr_ima += row_offset, + ptr_opacity_mask += opacity_row_offset) + for (def::coord col = mincol; col <= maxcol; ++col) + { + *pixels = impl::get_color(*ptr_ima++); + (*pixels).opacity = ((*ptr_opacity_mask++) ? 255 : 0); + ++pixels; + } + + view.sync(); + } + + } // end of namespace mln::io::magick::internal + + + template <typename I, typename J> void - save(const Image& ima_, const std::string& filename) + save(const Image& ima_, const Image<J>& opacity_mask_, + const std::string& filename) { trace::entering("mln::io::magick::save"); @@ -149,15 +247,13 @@ namespace mln } const I& ima = exact(ima_); + const J& opacity_mask = exact(opacity_mask_); def::coord - minrow = geom::min_row(ima), - mincol = geom::min_col(ima), - maxrow = geom::max_row(ima), - maxcol = geom::max_col(ima), ncols = geom::ncols(ima), nrows = geom::nrows(ima); + // In the construction of a Geometry object, the width (i.e. // `ncols') comes first, then the height (i.e. `nrows') // follows. @@ -168,28 +264,31 @@ namespace mln // declared further fails and segfault... Magick::Image magick_ima(Magick::Geometry(ncols, nrows), "white"); - magick_ima.type(Magick::TrueColorType); - - // Ensure that there is only one reference to underlying image - // If this is not done, then image pixels will not be modified. - magick_ima.modifyImage(); + if (opacity_mask.is_valid()) + { + magick_ima.type(Magick::TrueColorMatteType); + internal::paste_data_opacity(ima, opacity_mask, magick_ima); + } + else + { + magick_ima.type(Magick::TrueColorType); + internal::paste_data(ima, magick_ima); + } - Magick::Pixels view(magick_ima); - // As above, `ncols' is passed before `nrows'. - Magick::PixelPacket* pixels = view.get(0, 0, ncols, nrows); - const mln_value(I) *ptr_ima = &ima(ima.domain().pmin()); + magick_ima.write(filename); - unsigned row_offset = ima.delta_index(dpoint2d(+1, - ncols)); + trace::exiting("mln::io::magick::save"); + } - for (def::coord row = minrow; row <= maxrow; - ++row, ptr_ima += row_offset) - for (def::coord col = mincol; col <= maxcol; ++col) - *pixels++ = impl::get_color(*ptr_ima++); - view.sync(); - magick_ima.write(filename); - trace::exiting("mln::io::magick::save"); + template <typename I> + inline + void + save(const Image& ima, const std::string& filename) + { + mln_ch_value(I,bool) opacity_mask; + save(ima, opacity_mask, filename); } -- 1.5.6.5

1 0

last-svn-commit-856-g62776ae Small fixes in Scribo.
by Guillaume Lazzara 04 May '11

04 May '11

* demo/xml2doc/image_crop.cc: Abort xml loading if an error is found. * scribo/core/line_info.hh: Fix text cleanup. * scribo/io/img/save.hh: Add documentation. * scribo/io/xml/internal/extended_page_xml_visitor.hh: Use internal_gradient. * scribo/primitive/extract/alignments.hh: Set component type to whitespaceSeparator. * src/binarization/sauvola_ms_fg.cc: Fix argument use. --- scribo/ChangeLog | 19 +++++++++++++++++++ scribo/demo/xml2doc/image_crop.cc | 2 ++ scribo/scribo/core/line_info.hh | 3 ++- scribo/scribo/io/img/save.hh | 7 +++++++ .../io/xml/internal/extended_page_xml_visitor.hh | 4 ++-- scribo/scribo/primitive/extract/alignments.hh | 2 +- scribo/src/binarization/sauvola_ms_fg.cc | 6 ++++-- 7 files changed, 37 insertions(+), 6 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index fc761c3..a213078 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,24 @@ 2011-05-04 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Small fixes in Scribo. + + * demo/xml2doc/image_crop.cc: Abort xml loading if an error + is found. + + * scribo/core/line_info.hh: Fix text cleanup. + + * scribo/io/img/save.hh: Add documentation. + + * scribo/io/xml/internal/extended_page_xml_visitor.hh: Use + internal_gradient. + + * scribo/primitive/extract/alignments.hh: Set component type to + whitespaceSeparator. + + * src/binarization/sauvola_ms_fg.cc: Fix argument use. + +2011-05-04 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + Make paragraph extraction work with this branch. * scribo/core/line_info.hh: Keep few attributes public and update diff --git a/scribo/demo/xml2doc/image_crop.cc b/scribo/demo/xml2doc/image_crop.cc index 29a93d8..ba30d8a 100644 --- a/scribo/demo/xml2doc/image_crop.cc +++ b/scribo/demo/xml2doc/image_crop.cc @@ -326,6 +326,8 @@ bool ImageCrop::crop_regions(bool temp) else { f.close(); + qDebug() << "Error while reading XML file!"; + abort(); return false; } } diff --git a/scribo/scribo/core/line_info.hh b/scribo/scribo/core/line_info.hh index 27dda84..e91e1c3 100644 --- a/scribo/scribo/core/line_info.hh +++ b/scribo/scribo/core/line_info.hh @@ -351,7 +351,7 @@ namespace scribo static std::map<char, std::string> map = init_map(); std::string output = input; - for (unsigned i = 0; i < input.size(); ++i) + for (unsigned i = 0; i < output.size(); ++i) { std::map<char, std::string>::iterator it = map.find(output.at(i)); if (it != map.end()) @@ -360,6 +360,7 @@ namespace scribo i += it->second.size() - 1; } } + return output; } diff --git a/scribo/scribo/io/img/save.hh b/scribo/scribo/io/img/save.hh index 56300d6..04f0a3c 100644 --- a/scribo/scribo/io/img/save.hh +++ b/scribo/scribo/io/img/save.hh @@ -66,6 +66,13 @@ namespace scribo NonText : Display non-text regions. Full : Text and non-text regions. + + DebugWoImage : Display all regions onto a subsampled copy of + the original image. A dark background is used as background. + + DebugWithImage : Display all regions onto a subsampled copy of + the original image. The original image is used as background. + */ enum Format { diff --git a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh index b906a4a..b463677 100644 --- a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh +++ b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh @@ -32,7 +32,7 @@ # include <fstream> -# include <mln/morpho/elementary/gradient_external.hh> +# include <mln/morpho/elementary/gradient_internal.hh> # include <mln/pw/all.hh> # include <mln/core/image/dmorph/image_if.hh> @@ -122,7 +122,7 @@ namespace scribo if (doc.has_elements()) { // Prepare element edges - elt_edge = morpho::elementary::gradient_external(doc.elements().labeled_image(), c4()); + elt_edge = morpho::elementary::gradient_internal(doc.elements().labeled_image(), c4()); doc.elements().accept(*this); } diff --git a/scribo/scribo/primitive/extract/alignments.hh b/scribo/scribo/primitive/extract/alignments.hh index 8d18d93..1e3d835 100644 --- a/scribo/scribo/primitive/extract/alignments.hh +++ b/scribo/scribo/primitive/extract/alignments.hh @@ -1085,7 +1085,7 @@ namespace scribo V ndelim; component_set<L> - delim_comps = primitive::extract::components(delimitors, c8(), ndelim); + delim_comps = primitive::extract::components(delimitors, c8(), ndelim, component::WhitespaceSeparator); mln::util::couple<component_set<L>, mln_ch_value(L,bool)> output(delim_comps, delimitors); diff --git a/scribo/src/binarization/sauvola_ms_fg.cc b/scribo/src/binarization/sauvola_ms_fg.cc index 7ff9321..2c5e7fa 100644 --- a/scribo/src/binarization/sauvola_ms_fg.cc +++ b/scribo/src/binarization/sauvola_ms_fg.cc @@ -58,10 +58,10 @@ bool check_args(int argc, char * argv[]) 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)." }, + { "output.pbm", "A binary image." }, {0, 0} }; @@ -92,6 +92,8 @@ int main(int argc, char *argv[]) std::cout << "Using w_1=" << w_1 << " - s=" << s << std::endl; + Magick::InitializeMagick(0); + // Load image2d<value::rgb8> input_1; io::magick::load(input_1, argv[1]); @@ -108,7 +110,7 @@ int main(int argc, char *argv[]) image2d<bool> output = scribo::binarization::sauvola_ms(fg_gl, w_1, s, SCRIBO_DEFAULT_SAUVOLA_K); - io::pbm::save(output, argv[6]); + io::pbm::save(output, argv[5]); } -- 1.5.6.5

1 0

last-svn-commit-855-gae863e4 Make paragraph extraction work with this branch.
by Guillaume Lazzara 04 May '11

04 May '11

* scribo/core/line_info.hh: Keep few attributes public and update method names. * scribo/core/stats.hh: Fix missed elements in computation. * scribo/text/paragraphs.hh: Use line_set and line_links structures. * scribo/toolchain/internal/content_in_doc_functor.hh: Make use of the new paragraph extraction algorithm. --- scribo/ChangeLog | 15 + scribo/scribo/core/line_info.hh | 10 +- scribo/scribo/core/stats.hh | 13 +- scribo/scribo/text/paragraphs.hh | 1360 ++++++++++---------- .../toolchain/internal/content_in_doc_functor.hh | 178 ++-- 5 files changed, 808 insertions(+), 768 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index a6ef3f8..fc761c3 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,18 @@ +2011-05-04 Guillaume Lazzara <lazzara(a)fidji.lrde.epita.fr> + + Make paragraph extraction work with this branch. + + * scribo/core/line_info.hh: Keep few attributes public and update + method names. + + * scribo/core/stats.hh: Fix missed elements in computation. + + * scribo/text/paragraphs.hh: Use line_set and line_links + structures. + + * scribo/toolchain/internal/content_in_doc_functor.hh: Make use of + the new paragraph extraction algorithm. + 2011-05-03 Guillaume Lazzara <lazzara(a)lrde.epita.fr> * tests/unit_test/unit-tests.mk: Regen. diff --git a/scribo/scribo/core/line_info.hh b/scribo/scribo/core/line_info.hh index 949315a..27dda84 100644 --- a/scribo/scribo/core/line_info.hh +++ b/scribo/scribo/core/line_info.hh @@ -145,14 +145,12 @@ namespace scribo // Line set holding this element. line_set<L> holder_; - private: - void init_(); - // DEBUG stats< float > meanline_clusters_; stats< float > baseline_clusters_; - }; + private: + void init_(); }; } // end of namespace scribo::internal @@ -980,8 +978,8 @@ namespace scribo { const component_set<L>& comp_set = data_->holder_.components(); - const unsigned c1 = data_->components_(0); - const unsigned c2 = data_->components_(1); + const unsigned c1 = data_->component_ids_(0); + const unsigned c2 = data_->component_ids_(1); if (data_->holder_.components()(c1).type() == component::Punctuation || data_->holder_.components()(c2).type() == component::Punctuation) diff --git a/scribo/scribo/core/stats.hh b/scribo/scribo/core/stats.hh index bc24044..095735c 100644 --- a/scribo/scribo/core/stats.hh +++ b/scribo/scribo/core/stats.hh @@ -261,7 +261,7 @@ private: unsigned i = 0; const unsigned nelements = data_.nelements(); - clusters[0] = cluster_index; + clusters.push_back(cluster_index); const T std = data_.standard_deviation(); for (i = 1; i < nelements - 1; ++i) @@ -276,11 +276,12 @@ private: clusters.push_back(cluster_index); } - if (nelements > 1 - && data_[i] - data_[i - 1] > std) - ++cluster_index; - - clusters.push_back(cluster_index); + if (nelements > 1) + { + if (data_[i] - data_[i - 1] > std) + ++cluster_index; + clusters.push_back(cluster_index); + } clusters_.clear(); clusters_.reserve(cluster_index); diff --git a/scribo/scribo/text/paragraphs.hh b/scribo/scribo/text/paragraphs.hh index 5cb253e..94e5751 100644 --- a/scribo/scribo/text/paragraphs.hh +++ b/scribo/scribo/text/paragraphs.hh @@ -15,7 +15,9 @@ #include <scribo/core/macros.hh> #include <scribo/core/line_set.hh> +#include <scribo/core/line_links.hh> #include <scribo/core/line_info.hh> +#include <scribo/core/paragraph_set.hh> using namespace mln; @@ -59,15 +61,15 @@ namespace scribo template <typename L> inline - void paragraph_links(const util::array<value::int_u16>& left, - const util::array<value::int_u16>& right, - util::array<value::int_u16>& output, - const util::array< line_info<L> >& lines, + void paragraph_links(const line_links<L>& left, + const line_links<L>& right, + line_links<L>& output, + const line_set<L>& lines, const image2d<bool>& input) { - output = left; + output = left.duplicate(); - const unsigned nlines = lines.nelements(); + // const unsigned nlines = lines.nelements(); // image2d<value::rgb8> links = data::convert(value::rgb8(), input); // for (unsigned l = 0; l < nlines; ++l) @@ -77,242 +79,259 @@ namespace scribo // mln::io::ppm::save(links, "out_links.ppm"); // For each line - for (unsigned l = 0; l < nlines; ++l) - { - // Neighbors - - const value::int_u16 left_nbh = output(l); - const value::int_u16 right_nbh = right(l); - const value::int_u16 lol_nbh = output(left_nbh); - - // Line features - const float x_height = lines(l).x_height(); - const float left_x_height = lines(left_nbh).x_height(); - const float right_x_height = lines(right_nbh).x_height(); - - const box2d& left_line_bbox = lines(left_nbh).bbox(); - const box2d& current_line_bbox = lines(l).bbox(); - const box2d& right_line_bbox = lines(right_nbh).bbox(); - const box2d& lol_line_bbox = lines(lol_nbh).bbox(); // lol : left neighbor of the left neighbor - - const int lline_col_min = left_line_bbox.pmin().col(); - const int cline_col_min = current_line_bbox.pmin().col(); - const int rline_col_min = right_line_bbox.pmin().col(); - const int lolline_col_min = lol_line_bbox.pmin().col(); - - const int lline_col_max = left_line_bbox.pmax().col(); - const int cline_col_max = current_line_bbox.pmax().col(); - const int rline_col_max = right_line_bbox.pmax().col(); - - const int lline_cw = lines(left_nbh).char_width(); - const int cline_cw = lines(l).char_width(); - const int rline_cw = lines(right_nbh).char_width(); - // Maximal x variation to consider two lines vertically aligned - const int delta_alignment = cline_cw; - - // Checks the baseline distances of the two neighbors + for_all_lines(l, lines) + if (lines(l).is_textline()) { - // Current line baseline - const int c_baseline = lines(l).baseline(); - - // Baseline distance with the left and right neighbors - const int lc_baseline = lines(left_nbh).baseline() - c_baseline; - const int rc_baseline = c_baseline -lines(right_nbh).baseline(); + // Neighbors + + const line_id_t left_nbh = output(l); + const line_id_t right_nbh = right(l); + const line_id_t lol_nbh = output(left_nbh); + + // Line features + const float x_height = lines(l).x_height(); + const float left_x_height = lines(left_nbh).x_height(); + const float right_x_height = lines(right_nbh).x_height(); + + const box2d& left_line_bbox = lines(left_nbh).bbox(); + const box2d& current_line_bbox = lines(l).bbox(); + const box2d& right_line_bbox = lines(right_nbh).bbox(); + const box2d& lol_line_bbox = lines(lol_nbh).bbox(); // lol : left neighbor of the left neighbor + + const int lline_col_min = left_line_bbox.pmin().col(); + const int cline_col_min = current_line_bbox.pmin().col(); + const int rline_col_min = right_line_bbox.pmin().col(); + const int lolline_col_min = lol_line_bbox.pmin().col(); + + const int lline_col_max = left_line_bbox.pmax().col(); + const int cline_col_max = current_line_bbox.pmax().col(); + const int rline_col_max = right_line_bbox.pmax().col(); + + const int lline_cw = lines(left_nbh).char_width(); + const int cline_cw = lines(l).char_width(); + const int rline_cw = lines(right_nbh).char_width(); + // Maximal x variation to consider two lines vertically aligned + const int delta_alignment = cline_cw; + + // Checks the baseline distances of the two neighbors + { + // Current line baseline + const int c_baseline = lines(l).baseline(); - // Max baseline distance between the two neighbors - // const float delta_baseline_max = std::max(lc_baseline, rc_baseline); - // const float delta_baseline_min = std::min(lc_baseline, - // rc_baseline); + // Baseline distance with the left and right neighbors + const int lc_baseline = lines(left_nbh).baseline() - c_baseline; + const int rc_baseline = c_baseline -lines(right_nbh).baseline(); - // Only two lines, meaning the current line has only one neighbor - bool two_lines = false; + // Max baseline distance between the two neighbors + // const float delta_baseline_max = std::max(lc_baseline, rc_baseline); + // const float delta_baseline_min = std::min(lc_baseline, + // rc_baseline); - // If the current line has no left neighbor - if (lc_baseline == 0) - { - // ror : right neighbor of the right neighbor - const value::int_u16 ror_nbh = right(right_nbh); - // const box2d& ror_line_bbox = lines(ror_nbh).bbox(); + // Only two lines, meaning the current line has only one neighbor + bool two_lines = false; - // If the current line has a ror - if (ror_nbh != right_nbh - && output(ror_nbh) == right_nbh) + // If the current line has no left neighbor + if (lc_baseline == 0) { - // Distance between the current line and the right neighbor - const float right_distance = lines(l).meanline() - lines(right_nbh).baseline(); - // Distance between the right neighbor and the ror - const float ror_distance = lines(right_nbh).meanline() - lines(ror_nbh).baseline(); - // ror x_height - const float ror_x_height = lines(ror_nbh).x_height(); - - // Conditions to cut the link between the current line - // and its right neighbor - if (right_distance > 1.4f * ror_distance - && std::max(ror_x_height, right_x_height) < - 1.2f * std::min(ror_x_height, right_x_height) - && output(right_nbh) == l) + // ror : right neighbor of the right neighbor + const line_id_t ror_nbh = right(right_nbh); + //const box2d& ror_line_bbox = lines(ror_nbh).bbox(); + + // If the current line has a ror + if (ror_nbh != right_nbh + && output(ror_nbh) == right_nbh) { + // Distance between the current line and the right neighbor + const float right_distance = lines(l).meanline() - lines(right_nbh).baseline(); + // Distance between the right neighbor and the ror + const float ror_distance = lines(right_nbh).meanline() - lines(ror_nbh).baseline(); + // ror x_height + const float ror_x_height = lines(ror_nbh).x_height(); + + // Conditions to cut the link between the current line + // and its right neighbor + if (right_distance > 1.4f * ror_distance + && std::max(ror_x_height, right_x_height) < + 1.2f * std::min(ror_x_height, right_x_height) + && output(right_nbh) == l) + { output(right_nbh) = right_nbh; continue; + } } - } - // Otherwise we only have a group of two lines - else - { - // We determine the distance between the two lines - const float distance = lines(l).meanline() - lines(right_nbh).baseline(); - two_lines = true; - - // If the distance between the two lines is greater than - // the minimum x height of the two lines then we cut the - // link between them - if (distance > 2.0f * std::min(x_height, right_x_height) - && output(right_nbh) == l) + // Otherwise we only have a group of two lines + else { + // We determine the distance between the two lines + const float distance = lines(l).meanline() - lines(right_nbh).baseline(); + two_lines = true; + + // If the distance between the two lines is greater than + // the minimum x height of the two lines then we cut the + // link between them + if (distance > 2.0f * std::min(x_height, right_x_height) + && output(right_nbh) == l) + { output(right_nbh) = right_nbh; continue; + } } - } - // Lines features - const float min_x_height = std::min(x_height, right_x_height); - const float max_x_height = std::max(x_height, right_x_height); - const float min_char_width = std::min(rline_cw, cline_cw); - const float max_char_width = std::max(rline_cw, cline_cw); + // Lines features + const float min_x_height = std::min(x_height, right_x_height); + const float max_x_height = std::max(x_height, right_x_height); + const float min_char_width = std::min(rline_cw, cline_cw); + const float max_char_width = std::max(rline_cw, cline_cw); - // Condition to cut the link between the current line and - // its right neighbor - if ((max_x_height > min_x_height * 1.2f) && - !(max_char_width <= 1.2f * min_char_width)) - { - if (output(right_nbh) == l) + // Condition to cut the link between the current line and + // its right neighbor + if ((max_x_height > min_x_height * 1.2f) && + !(max_char_width <= 1.2f * min_char_width)) { - output(right_nbh) = right_nbh; - continue; + if (output(right_nbh) == l) + { + output(right_nbh) = right_nbh; + continue; + } } - } - - // If we only have two lines we stop the study - if (two_lines) - continue; - } - // If the current line has no right neighbor - else if (rc_baseline == 0) - { - // lol : left neighbor of the left neighbor - // If the left neighbor of the current line has a left neighbor - if (lol_nbh != left_nbh) + // If we only have two lines we stop the study + if (two_lines) + continue; + } + // If the current line has no right neighbor + else if (rc_baseline == 0) { - // Distance between the current line and its left neighbor - const float left_distance = lines(left_nbh).meanline() - - lines(l).baseline(); - // Distance between the left neighbor and the left - // neighbor of its left neighbor - const float lol_distance = lines(lol_nbh).meanline() - - lines(left_nbh).baseline(); - // lol x height - const float lol_x_height = lines(lol_nbh).x_height(); - - // Conditions to cut the link between the current line - // and its left neighbor - if (left_distance > 1.4f * lol_distance - && std::max(lol_x_height, left_x_height) < - 1.2f * std::min(lol_x_height, left_x_height)) + // lol : left neighbor of the left neighbor + + // If the left neighbor of the current line has a left neighbor + if (lol_nbh != left_nbh) { + // Distance between the current line and its left neighbor + const float left_distance = lines(left_nbh).meanline() - + lines(l).baseline(); + // Distance between the left neighbor and the left + // neighbor of its left neighbor + const float lol_distance = lines(lol_nbh).meanline() - + lines(left_nbh).baseline(); + // lol x height + const float lol_x_height = lines(lol_nbh).x_height(); + + // Conditions to cut the link between the current line + // and its left neighbor + if (left_distance > 1.4f * lol_distance + && std::max(lol_x_height, left_x_height) < + 1.2f * std::min(lol_x_height, left_x_height)) + { output(l) = l; continue; + } } - } - // Otherwise we only have a group of two lines - else - { - // Distance between the current line and it left neighbor - const float distance = lines(left_nbh).meanline() - - lines(l).baseline(); + // Otherwise we only have a group of two lines + else + { + // Distance between the current line and it left neighbor + const float distance = lines(left_nbh).meanline() - + lines(l).baseline(); - two_lines = true; + two_lines = true; - // If the distance is greater than the min x height - // between the two lines - if (distance > 2.0f * std::min(x_height, left_x_height)) - { + // If the distance is greater than the min x height + // between the two lines + if (distance > 2.0f * std::min(x_height, left_x_height)) + { output(l) = l; continue; + } } - } - - // Lines features - const float min_x_height = std::min(x_height, left_x_height); - const float max_x_height = std::max(x_height, left_x_height); - const float min_char_width = std::min(lline_cw, cline_cw); - const float max_char_width = std::max(lline_cw, cline_cw); - - // Condition to cut the link between the current line and - // its left neighbor - if ((max_x_height > min_x_height * 1.2f) && - !(max_char_width <= 1.2f * min_char_width)) - { - output(l) = l; - continue; - } - // If we only have two lines we stop the study - if (two_lines) - continue; - } - // The current line has at least one left and one right neighbor - else // if (delta_baseline_max >= delta_baseline_min) - { - // Distance between the left and the current line - const float left_distance = - lines(left_nbh).meanline() - lines(l).baseline(); - // Distance between the right and the current line - const float right_distance = - lines(l).meanline() - lines(right_nbh).baseline(); - - // If the left line is too far compared to the right one - // we cut the link with it - if (left_distance > 1.2f * right_distance - && std::max(x_height, left_x_height) > 1.2f * std::min(x_height, left_x_height)) - { - output(l) = l; - continue; - } - // If the right line is too far compared to the left one - // we cut the link with it - else if (right_distance > 1.2f * left_distance - && std::max(x_height, right_x_height) > 1.2f * std::min(x_height, right_x_height) - && output(right_nbh) == l) - { - output(right_nbh) = right_nbh; - continue; - } - - // If the distance between the baseline of the left - // neighbor and the baseline of the current line is - // greater than the one between the current line baseline - // and the right line baseline we have to study the texte - // features of the right and left lines - if (lc_baseline > rc_baseline) - { - const float cw_max = std::max(lline_cw, cline_cw); - const float cw_min = std::min(lline_cw, cline_cw); + // Lines features const float min_x_height = std::min(x_height, left_x_height); const float max_x_height = std::max(x_height, left_x_height); + const float min_char_width = std::min(lline_cw, cline_cw); + const float max_char_width = std::max(lline_cw, cline_cw); + // Condition to cut the link between the current line and + // its left neighbor if ((max_x_height > min_x_height * 1.2f) && - !(cw_max <= 1.2f * cw_min)) + !(max_char_width <= 1.2f * min_char_width)) { output(l) = l; continue; } + // If we only have two lines we stop the study + if (two_lines) + continue; + } + // The current line has at least one left and one right neighbor + else // if (delta_baseline_max >= delta_baseline_min) + { + // Distance between the left and the current line + const float left_distance = + lines(left_nbh).meanline() - lines(l).baseline(); + // Distance between the right and the current line + const float right_distance = + lines(l).meanline() - lines(right_nbh).baseline(); + + // If the left line is too far compared to the right one + // we cut the link with it + if (left_distance > 1.2f * right_distance + && std::max(x_height, left_x_height) > 1.2f * std::min(x_height, left_x_height)) + { + output(l) = l; + continue; + } + // If the right line is too far compared to the left one + // we cut the link with it + else if (right_distance > 1.2f * left_distance + && std::max(x_height, right_x_height) > 1.2f * std::min(x_height, right_x_height) + && output(right_nbh) == l) + { + output(right_nbh) = right_nbh; + continue; + } + + // If the distance between the baseline of the left + // neighbor and the baseline of the current line is + // greater than the one between the current line baseline + // and the right line baseline we have to study the texte + // features of the right and left lines + if (lc_baseline > rc_baseline) + { + const float cw_max = std::max(lline_cw, cline_cw); + const float cw_min = std::min(lline_cw, cline_cw); + const float min_x_height = std::min(x_height, left_x_height); + const float max_x_height = std::max(x_height, left_x_height); + + if ((max_x_height > min_x_height * 1.2f) && + !(cw_max <= 1.2f * cw_min)) + { + output(l) = l; + continue; + } + + { + const float min_x_height = std::min(x_height, right_x_height); + const float max_x_height = std::max(x_height, right_x_height); + const float cw_max = std::max(rline_cw, cline_cw); + const float cw_min = std::min(rline_cw, cline_cw); + + if ((max_x_height > min_x_height * 1.2f) + && !(cw_max <= 1.2f * cw_min) + && output(right_nbh) == l) + { + output(right_nbh) = right_nbh; + continue; + } + } + } + else { - const float min_x_height = std::min(x_height, right_x_height); - const float max_x_height = std::max(x_height, right_x_height); const float cw_max = std::max(rline_cw, cline_cw); const float cw_min = std::min(rline_cw, cline_cw); + const float min_x_height = std::min(x_height, right_x_height); + const float max_x_height = std::max(x_height, right_x_height); if ((max_x_height > min_x_height * 1.2f) && !(cw_max <= 1.2f * cw_min) @@ -321,44 +340,28 @@ namespace scribo output(right_nbh) = right_nbh; continue; } - } - } - else - { - const float cw_max = std::max(rline_cw, cline_cw); - const float cw_min = std::min(rline_cw, cline_cw); - const float min_x_height = std::min(x_height, right_x_height); - const float max_x_height = std::max(x_height, right_x_height); - if ((max_x_height > min_x_height * 1.2f) - && !(cw_max <= 1.2f * cw_min) - && output(right_nbh) == l) - { - output(right_nbh) = right_nbh; - continue; - } - - { - const float min_x_height = std::min(x_height, left_x_height); - const float max_x_height = std::max(x_height, left_x_height); - const float cw_max = std::max(lline_cw, cline_cw); - const float cw_min = std::min(lline_cw, cline_cw); - - if ((max_x_height > min_x_height * 1.2f) - && !(cw_max <= 1.2f * cw_min)) { - output(l) = l; - continue; + const float min_x_height = std::min(x_height, left_x_height); + const float max_x_height = std::max(x_height, left_x_height); + const float cw_max = std::max(lline_cw, cline_cw); + const float cw_min = std::min(lline_cw, cline_cw); + + if ((max_x_height > min_x_height * 1.2f) + && !(cw_max <= 1.2f * cw_min)) + { + output(l) = l; + continue; + } } } } } - } - // If we arrive here, it means than the lines in the - // neighborhood of the current line are quite similar. We can - // then begin to study the indentations in order to determine - // the beginning of new paragraphs + // If we arrive here, it means than the lines in the + // neighborhood of the current line are quite similar. We can + // then begin to study the indentations in order to determine + // the beginning of new paragraphs //----------------------------------------------------------------------------- // ___________________________ @@ -375,35 +378,35 @@ namespace scribo // //----------------------------------------------------------------------------- - { - // Check if the current line neighbors are aligned - bool left_right_aligned = false; - bool left_lol_aligned = false; - const int dx_lr = std::abs(lline_col_min - rline_col_min); - const int dx_llol = std::abs(lline_col_min - lolline_col_min); + { + // Check if the current line neighbors are aligned + bool left_right_aligned = false; + bool left_lol_aligned = false; + const int dx_lr = std::abs(lline_col_min - rline_col_min); + const int dx_llol = std::abs(lline_col_min - lolline_col_min); - if (dx_lr < delta_alignment) - left_right_aligned = true; + if (dx_lr < delta_alignment) + left_right_aligned = true; - if (dx_llol < delta_alignment) - left_lol_aligned = true; + if (dx_llol < delta_alignment) + left_lol_aligned = true; - if (left_right_aligned && left_lol_aligned) - { - const int left_right_col_min = std::min(lline_col_min, rline_col_min); - const int dx_lrc = std::abs(left_right_col_min - cline_col_min); - const float l_char_width = 1.5f * lines(l).char_width(); - - if (dx_lrc > l_char_width && - dx_lrc < 3.0f * l_char_width && - cline_col_min > rline_col_min && - cline_col_min > lline_col_min) + if (left_right_aligned && left_lol_aligned) { - output(right_nbh) = right_nbh; - continue; + const int left_right_col_min = std::min(lline_col_min, rline_col_min); + const int dx_lrc = std::abs(left_right_col_min - cline_col_min); + const float l_char_width = 1.5f * lines(l).char_width(); + + if (dx_lrc > l_char_width && + dx_lrc < 3.0f * l_char_width && + cline_col_min > rline_col_min && + cline_col_min > lline_col_min) + { + output(right_nbh) = right_nbh; + continue; + } } } - } //----------------------------------------------------------------------------- // ___________________________ @@ -419,41 +422,41 @@ namespace scribo // //----------------------------------------------------------------------------- - { - // Check if the current line neighbors are aligned - bool left_right_max_aligned = false; - bool left_current_min_aligned = false; - bool lol_current_min_aligned = false; - const bool lol_is_left = output(left_nbh) == left_nbh; - const int dx_lr_max = std::abs(lline_col_max - rline_col_max); - const int dx_lc_min = std::abs(lline_col_min - cline_col_min); - const int dx_lolc_min = std::abs(lolline_col_min - cline_col_min); - - if (dx_lr_max < delta_alignment) - left_right_max_aligned = true; - - if (dx_lc_min < delta_alignment) - left_current_min_aligned = true; - - if (dx_lolc_min < delta_alignment) - lol_current_min_aligned = true; - - if (!left_current_min_aligned && left_right_max_aligned && - (lol_current_min_aligned || lol_is_left)) { - const int dx_lrc = std::abs(lline_col_max - cline_col_max); - const int l_char_width = lines(l).char_width(); - - if (dx_lrc > l_char_width && - cline_col_max < lline_col_max && - cline_col_min < lline_col_min && - (lline_col_min > lolline_col_min || lol_is_left)) + // Check if the current line neighbors are aligned + bool left_right_max_aligned = false; + bool left_current_min_aligned = false; + bool lol_current_min_aligned = false; + const bool lol_is_left = output(left_nbh) == left_nbh; + const int dx_lr_max = std::abs(lline_col_max - rline_col_max); + const int dx_lc_min = std::abs(lline_col_min - cline_col_min); + const int dx_lolc_min = std::abs(lolline_col_min - cline_col_min); + + if (dx_lr_max < delta_alignment) + left_right_max_aligned = true; + + if (dx_lc_min < delta_alignment) + left_current_min_aligned = true; + + if (dx_lolc_min < delta_alignment) + lol_current_min_aligned = true; + + if (!left_current_min_aligned && left_right_max_aligned && + (lol_current_min_aligned || lol_is_left)) { - output(l) = l; - continue; + const int dx_lrc = std::abs(lline_col_max - cline_col_max); + const int l_char_width = lines(l).char_width(); + + if (dx_lrc > l_char_width && + cline_col_max < lline_col_max && + cline_col_min < lline_col_min && + (lline_col_min > lolline_col_min || lol_is_left)) + { + output(l) = l; + continue; + } } } - } //----------------------------------------------------------------------------- @@ -469,68 +472,69 @@ namespace scribo // //----------------------------------------------------------------------------- - if (left_nbh == l) - { - const value::int_u16 ror_nbh = right(right_nbh); - const box2d& ror_line_bbox = lines(ror_nbh).bbox(); - const int rorline_col_min = ror_line_bbox.pmin().col(); + if (left_nbh == l) + { + const line_id_t ror_nbh = right(right_nbh); + const box2d& ror_line_bbox = lines(ror_nbh).bbox(); + const int rorline_col_min = ror_line_bbox.pmin().col(); - bool right_ror_min_aligned = false; - const int dx_rror_min = std::abs(rline_col_min - rorline_col_min); + bool right_ror_min_aligned = false; + const int dx_rror_min = std::abs(rline_col_min - rorline_col_min); - if (dx_rror_min < delta_alignment) - right_ror_min_aligned = true; + if (dx_rror_min < delta_alignment) + right_ror_min_aligned = true; - if (right_ror_min_aligned) - { - const int right_ror_col_min = std::min(rline_col_min, rorline_col_min); - const int dx_rrorc = std::abs(right_ror_col_min - cline_col_min); - const float l_char_width = 1.5f * lines(l).char_width(); - - if (dx_rrorc > l_char_width && - dx_rrorc < 3.0f * l_char_width && - cline_col_min > rline_col_min && - cline_col_max >= rline_col_max) + if (right_ror_min_aligned) { - output(right_nbh) = right_nbh; - continue; + const int right_ror_col_min = std::min(rline_col_min, rorline_col_min); + const int dx_rrorc = std::abs(right_ror_col_min - cline_col_min); + const float l_char_width = 1.5f * lines(l).char_width(); + + if (dx_rrorc > l_char_width && + dx_rrorc < 3.0f * l_char_width && + cline_col_min > rline_col_min && + cline_col_max >= rline_col_max) + { + output(right_nbh) = right_nbh; + continue; + } } } } - } // Only debug - { - image2d<value::rgb8> debug = data::convert(value::rgb8(), input); - - for (unsigned i = 0; i < output.nelements(); ++i) - output(i) = internal::find_root(output, i); - - mln::util::array<accu::shape::bbox<point2d> > nbbox(output.nelements()); - for (unsigned i = 0; i < nlines; ++i) - { - // if (lines(i).is_textline()) - // { - // mln::draw::box(debug, lines(i).bbox(), literal::red); - nbbox(output(i)).take(lines(i).bbox()); - // } - } - - for (unsigned i = 0; i < nbbox.nelements(); ++i) - if (nbbox(i).is_valid()) - { - box2d b = nbbox(i).to_result(); - mln::draw::box(debug, b, literal::orange); - b.enlarge(1); - mln::draw::box(debug, b, literal::orange); - b.enlarge(1); - mln::draw::box(debug, b, literal::orange); - } - - mln::io::ppm::save(debug, "out_paragraph.ppm"); - } + // { + // image2d<value::rgb8> debug = data::convert(value::rgb8(), input); + + // for (unsigned i = 0; i < output.nelements(); ++i) + // output(i) = scribo::make::internal::find_root(output, i); + + // mln::util::array<accu::shape::bbox<point2d> > nbbox(output.nelements()); + // for_all_lines(l, lines) + // if (lines(l).is_textline()) + // { + // // if (lines(i).is_textline()) + // // { + // // mln::draw::box(debug, lines(i).bbox(), literal::red); + // nbbox(output(l)).take(lines(l).bbox()); + // // } + // } + + // for (unsigned i = 0; i < nbbox.nelements(); ++i) + // if (nbbox(i).is_valid()) + // { + // box2d b = nbbox(i).to_result(); + // mln::draw::box(debug, b, literal::orange); + // b.enlarge(1); + // mln::draw::box(debug, b, literal::orange); + // b.enlarge(1); + // mln::draw::box(debug, b, literal::orange); + // } + + // mln::io::ppm::save(debug, "out_paragraph.ppm"); + // } } } @@ -547,262 +551,269 @@ namespace scribo template <typename L> inline void prepare_lines(const box2d& domain, - const util::array< line_info<L> >& lines, - image2d<value::int_u16>& blocks, + const line_set<L>& lines, + L& blocks, util::array<box2d>& rbbox) { std::map< int, std::vector< const box2d* > > drawn_lines; - const unsigned nlines = lines.nelements(); + // const unsigned nlines = lines.nelements(); // For each line - for (unsigned l = 0; l < nlines; ++l) - { - // Rotation of the bounding box - box2d b = geom::rotate(lines(l).bbox(), -90, domain.pcenter()); - rbbox.append(b); - - const unsigned index = l + 1; - const unsigned even_index = 2 * index; - const unsigned odd_index = even_index + 1; - - // Top of the line + //for (unsigned l = 0; l < nlines; ++l) + for_all_lines(l, lines) + if (lines(l).is_textline()) { - bool not_finished = true; - int col_offset = 0; + // Rotation of the bounding box + box2d b = geom::rotate(lines(l).bbox(), -90, domain.pcenter()); +// rbbox.append(b); + rbbox(l) = b; - while (not_finished) - { - // Looking for a column in the image to draw the top of the - // line + const unsigned index = l + 1; + const unsigned even_index = 2 * index; + const unsigned odd_index = even_index + 1; - const int col = b.pmax().col() + col_offset; - std::map< int, std::vector< const box2d* > >::iterator it - = drawn_lines.find(col); + // Top of the line + { + bool not_finished = true; + int col_offset = 0; - if (it != drawn_lines.end()) + while (not_finished) { - const std::vector< const box2d* >& lines = (*it).second; - const unsigned nb_lines = lines.size(); - unsigned i = 0; + // Looking for a column in the image to draw the top of the + // line + + const int col = b.pmax().col() + col_offset; + std::map< int, std::vector< const box2d* > >::iterator it + = drawn_lines.find(col); - for (i = 0; i < nb_lines; ++i) + if (it != drawn_lines.end()) { - const box2d* box = lines[i]; - const int min_row = std::max(b.pmin().row(), box->pmin().row()); - const int max_row = std::min(b.pmax().row(), box->pmax().row()); + const std::vector< const box2d* >& lines = (*it).second; + const unsigned nb_lines = lines.size(); + unsigned i = 0; - if (min_row - max_row <= 0) - break; - } + for (i = 0; i < nb_lines; ++i) + { + const box2d* box = lines[i]; + const int min_row = std::max(b.pmin().row(), box->pmin().row()); + const int max_row = std::min(b.pmax().row(), box->pmax().row()); - if (i == nb_lines) + if (min_row - max_row <= 0) + break; + } + + if (i == nb_lines) + { + mln::draw::line(blocks, point2d(b.pmin().row(), col), + point2d(b.pmax().row(), col), even_index); + not_finished = false; + drawn_lines[col].push_back(&(rbbox[l])); + } + else + ++col_offset; + } + else { mln::draw::line(blocks, point2d(b.pmin().row(), col), point2d(b.pmax().row(), col), even_index); not_finished = false; drawn_lines[col].push_back(&(rbbox[l])); } - else - ++col_offset; - } - else - { - mln::draw::line(blocks, point2d(b.pmin().row(), col), - point2d(b.pmax().row(), col), even_index); - not_finished = false; - drawn_lines[col].push_back(&(rbbox[l])); } } - } - // Bottom of the line - { - bool not_finished = true; - int col_offset = 0; - - while (not_finished) + // Bottom of the line { - // Looking for a column in the image to draw the bottom of - // the line + bool not_finished = true; + int col_offset = 0; - const int col = b.pmin().col() - col_offset; - std::map< int, std::vector< const box2d* > >::iterator it - = drawn_lines.find(col); - - if (it != drawn_lines.end()) + while (not_finished) { - const std::vector< const box2d* >& lines = (*it).second; - const unsigned nb_lines = lines.size(); - unsigned i = 0; + // Looking for a column in the image to draw the bottom of + // the line + + const int col = b.pmin().col() - col_offset; + std::map< int, std::vector< const box2d* > >::iterator it + = drawn_lines.find(col); - for (i = 0; i < nb_lines; ++i) + if (it != drawn_lines.end()) { - const box2d* box = lines[i]; - const int min_row = std::max(b.pmin().row(), box->pmin().row()); - const int max_row = std::min(b.pmax().row(), box->pmax().row()); + const std::vector< const box2d* >& lines = (*it).second; + const unsigned nb_lines = lines.size(); + unsigned i = 0; - if (min_row - max_row <= 0) - break; - } + for (i = 0; i < nb_lines; ++i) + { + const box2d* box = lines[i]; + const int min_row = std::max(b.pmin().row(), box->pmin().row()); + const int max_row = std::min(b.pmax().row(), box->pmax().row()); - if (i == nb_lines) + if (min_row - max_row <= 0) + break; + } + + if (i == nb_lines) + { + mln::draw::line(blocks, point2d(b.pmin().row(), col), + point2d(b.pmax().row(), col), odd_index); + not_finished = false; + drawn_lines[col].push_back(&(rbbox[l])); + } + else + ++col_offset; + } + else { mln::draw::line(blocks, point2d(b.pmin().row(), col), point2d(b.pmax().row(), col), odd_index); not_finished = false; drawn_lines[col].push_back(&(rbbox[l])); } - else - ++col_offset; - } - else - { - mln::draw::line(blocks, point2d(b.pmin().row(), col), - point2d(b.pmax().row(), col), odd_index); - not_finished = false; - drawn_lines[col].push_back(&(rbbox[l])); } } } - } } template <typename L> inline void - process_left_link(image2d<value::int_u16>& blocks, + process_left_link(L& blocks, const util::array<box2d>& rbbox, - const util::array< line_info<L> >& lines, - util::array<value::int_u16>& left) + const line_set<L>& lines, + line_links<L>& left) { - typedef value::int_u16 V; + typedef scribo::def::lbl_type V; // At the beginning each line is its own neighbor - for (unsigned i = 0; i < left.nelements(); ++i) - left(i) = i; + for_all_lines(l, lines) + if (lines(l).is_textline()) + left(l) = l; + else + left(l) = 0; - const unsigned nlines = lines.nelements(); + // const unsigned nlines = lines.nelements(); // For each line - for (unsigned i = 0; i < nlines; ++i) - { - // Max distance for the line search - int dmax = 1.5f * lines(i).x_height(); - - // Starting points in the current line box - point2d c = rbbox(i).pcenter(); - point2d q(rbbox(i).pmin().row() + ((c.row() - rbbox(i).pmin().row()) / 4), c.col()); + for_all_lines(l, lines) + if (lines(l).is_textline()) + { + // Max distance for the line search + int dmax = 1.5f * lines(l).x_height(); - int - midcol = (rbbox(i).pmax().col() - - rbbox(i).pmin().col()) / 2; + // Starting points in the current line box + point2d c = rbbox(l).pcenter(); + point2d q(rbbox(l).pmin().row() + ((c.row() - rbbox(l).pmin().row()) / 4), c.col()); - // Left - { - // marge gauche int - nleftima = c.col() - blocks.domain().pmin().col(), - // Distance gauche - nleft = std::min(nleftima, midcol + dmax); - - V - // Starting points in the box - *p = &blocks(c), - *p2 = &blocks(q), - // End of search - *pstop = p - nleft - 1, - // Line neighbor - *nbh_p = 0; - - // While we haven't found a neighbor or reached the limit - for (; p != pstop; --p, --p2) + midcol = (rbbox(l).pmax().col() + - rbbox(l).pmin().col()) / 2; + + // Left { - if (*p2 != literal::zero // Not the background - && ((*p2 % 2) == 0) // Looking for the bottom of a line - && left((*p2 >> 1) - 1) != i) // No loops + // marge gauche + int + nleftima = c.col() - blocks.domain().pmin().col(), + // Distance gauche + nleft = std::min(nleftima, midcol + dmax); + + V + // Starting points in the box + *p = &blocks(c), + *p2 = &blocks(q), + // End of search + *pstop = p - nleft - 1, + // Line neighbor + *nbh_p = 0; + + // While we haven't found a neighbor or reached the limit + for (; p != pstop; --p, --p2) { - // Neightbor found, we stop the research - nbh_p = p2; - break; - } + if (*p2 != literal::zero // Not the background + && ((*p2 % 2) == 0) // Looking for the bottom of a line + && left((*p2 >> 1) - 1) != l) // No loops + { + // Neightbor found, we stop the research + nbh_p = p2; + break; + } - if (*p != literal::zero // Not the background - && ((*p % 2) == 0) // Looking for the bottom of a line - && left((*p >> 1) - 1) != i) // No loops - { - // Neightbor found, we stop the research - nbh_p = p; - break; + if (*p != literal::zero // Not the background + && ((*p % 2) == 0) // Looking for the bottom of a line + && left((*p >> 1) - 1) != l) // No loops + { + // Neightbor found, we stop the research + nbh_p = p; + break; + } } - } - - // If a neighbor was found, then we have found the top of the - // line. We are then looking for the bottom of the encountered - // line. If during the search process we find a complete line - // included in the touched line, this line is considered as - // the neighbor under certain conditions (see below) - - //--------------------------------------------------------------- - // _________________________ | - // |_________________________| => Current line | Search direction - // v - // => First encountered top line - // __________________________________________________ 2Q - // | Q | - // | _________________________ |2P - // | |_____________P___________| => Second top |2P + 1 - // | line | - // |__________________________________________________|2Q + 1 - // - // - //--------------------------------------------------------------- - - if (nbh_p) - { - std::vector<V> lines_nbh; - const V end_p = *nbh_p + 1; - const V* nbh_p_copy = nbh_p; - for (; *nbh_p != end_p; --nbh_p) + // If a neighbor was found, then we have found the top of the + // line. We are then looking for the bottom of the encountered + // line. If during the search process we find a complete line + // included in the touched line, this line is considered as + // the neighbor under certain conditions (see below) + + //--------------------------------------------------------------- + // _________________________ | + // |_________________________| => Current line | Search direction + // v + // => First encountered top line + // __________________________________________________ 2Q + // | Q | + // | _________________________ |2P + // | |_____________P___________| => Second top |2P + 1 + // | line | + // |__________________________________________________|2Q + 1 + // + // + //--------------------------------------------------------------- + + if (nbh_p) { - if ((*nbh_p) != literal::zero) // Not the background + std::vector<V> lines_nbh; + const V end_p = *nbh_p + 1; + const V* nbh_p_copy = nbh_p; + + for (; *nbh_p != end_p; --nbh_p) { - if ((*nbh_p) % 2 == 0)// We have found the top of - // another line - lines_nbh.push_back(*nbh_p); - else + if ((*nbh_p) != literal::zero) // Not the background { - // We have found the bottom of a line. We are looking if - // we have already encountered the top of this - // line. If so, we link the current line with this one - // under certain conditions: - - if (std::find(lines_nbh.begin(), lines_nbh.end(), - (*nbh_p) - 1) != lines_nbh.end()) + if ((*nbh_p) % 2 == 0)// We have found the top of + // another line + lines_nbh.push_back(*nbh_p); + else { - // If we can link the complete line with the current line - if (// It must be in the search range - nbh_p > pstop - // Avoid loops - && left(((*nbh_p - 1) >> 1) - 1) != i) - left(i) = ((*nbh_p - 1) >> 1) - 1; - - // We have found a complete line so we stop the search - break; + // We have found the bottom of a line. We are looking if + // we have already encountered the top of this + // line. If so, we link the current line with this one + // under certain conditions: + + if (std::find(lines_nbh.begin(), lines_nbh.end(), + (*nbh_p) - 1) != lines_nbh.end()) + { + // If we can link the complete line with the current line + if (// It must be in the search range + nbh_p > pstop + // Avoid loops + && left(((*nbh_p - 1) >> 1) - 1) != l) + left(l) = ((*nbh_p - 1) >> 1) - 1; + + // We have found a complete line so we stop the search + break; + } } } } - } - // If we haven't found any included line in the first - // neighbor, then the line is considered as the neighbor of - // the current line - if (*nbh_p == end_p) - left(i) = (*nbh_p_copy >> 1) - 1; + // If we haven't found any included line in the first + // neighbor, then the line is considered as the neighbor of + // the current line + if (*nbh_p == end_p) + left(l) = (*nbh_p_copy >> 1) - 1; + } } } - } } @@ -810,137 +821,141 @@ namespace scribo template <typename L> inline void - process_right_link(image2d<value::int_u16>& blocks, + process_right_link(L& blocks, const util::array<box2d>& rbbox, - const util::array< line_info<L> >& lines, - util::array<value::int_u16>& right) + const line_set<L>& lines, + line_links<L>& right) { - typedef value::int_u16 V; + typedef scribo::def::lbl_type V; // At the beginning each line is its own neighbor - for (unsigned i = 0; i < right.nelements(); ++i) - right(i) = i; + for_all_lines(l, lines) + if (lines(l).is_textline()) + right(l) = l; + else + right(l) = 0; - const unsigned nlines = lines.nelements(); + // const unsigned nlines = lines.nelements(); // For each line - for (unsigned i = 0; i < nlines; ++i) - { - // Max distance for the line search - int dmax = 1.5f * lines(i).x_height(); - - // Starting points in the current line box - point2d c = rbbox(i).pcenter(); - point2d q(rbbox(i).pmax().row() - ((rbbox(i).pmax().row() - c.row()) / 4), c.col()); + for_all_lines(l, lines) + if (lines(l).is_textline()) + { + // Max distance for the line search + int dmax = 1.5f * lines(l).x_height(); - int - midcol = (rbbox(i).pmax().col() - - rbbox(i).pmin().col()) / 2; + // Starting points in the current line box + point2d c = rbbox(l).pcenter(); + point2d q(rbbox(l).pmax().row() - ((rbbox(l).pmax().row() - c.row()) / 4), c.col()); - // Right - { int - nrightima = geom::ncols(blocks) - c.col() + blocks.domain().pmin().col(), - nright = std::min(nrightima, midcol + dmax); - - V - // Starting points in the box - *p = &blocks(c), - *p2 = &blocks(q), - // End of search - *pstop = p + nright - 1, - // Line neighbor - *nbh_p = 0; - - // While we haven't found a neighbor or reached the limit - for (; p != pstop; ++p, ++p2) + midcol = (rbbox(l).pmax().col() + - rbbox(l).pmin().col()) / 2; + + // Right { - if (*p2 != literal::zero // Not the background - && ((*p2 % 2) == 1) // Looking for the bottom of a line - && right(((*p2 - 1) >> 1) - 1) != i) // No loops + int + nrightima = geom::ncols(blocks) - c.col() + blocks.domain().pmin().col(), + nright = std::min(nrightima, midcol + dmax); + + V + // Starting points in the box + *p = &blocks(c), + *p2 = &blocks(q), + // End of search + *pstop = p + nright - 1, + // Line neighbor + *nbh_p = 0; + + // While we haven't found a neighbor or reached the limit + for (; p != pstop; ++p, ++p2) { - // Neightbor found, we stop the research - nbh_p = p2; - break; - } + if (*p2 != literal::zero // Not the background + && ((*p2 % 2) == 1) // Looking for the bottom of a line + && right(((*p2 - 1) >> 1) - 1) != l) // No loops + { + // Neightbor found, we stop the research + nbh_p = p2; + break; + } - if (*p != literal::zero // Not the background - && ((*p % 2) == 1) // Looking for the bottom of a line - && right(((*p - 1) >> 1) - 1) != i) // No loops - { - // Neightbor found, we stop the research - nbh_p = p; - break; + if (*p != literal::zero // Not the background + && ((*p % 2) == 1) // Looking for the bottom of a line + && right(((*p - 1) >> 1) - 1) != l) // No loops + { + // Neightbor found, we stop the research + nbh_p = p; + break; + } } - } - - // If a neighbor was found, then we have found the bottom of the - // line. We are then looking for the top of the encountered - // line. If during the search process we find a complete line - // included in the touched line, this line is considered as - // the neighbor under certain conditions (see below) - - //--------------------------------------------------------------- - // - // - // __________________________________________________ 2Q - // | Q | - // | _________________________ |2P - // | |_____________P___________| => Second bottom |2P + 1 - // | line | - // |__________________________________________________|2Q + 1 - // => First encountered bottom line - // _________________________ ^ - // |_________________________| => Current line | Search direction - // | - //--------------------------------------------------------------- - - if (nbh_p) - { - std::vector<V> lines_nbh; - const V end_p = *nbh_p - 1; - const V* nbh_p_copy = nbh_p; - for (; *nbh_p != end_p; ++nbh_p) + // If a neighbor was found, then we have found the bottom of the + // line. We are then looking for the top of the encountered + // line. If during the search process we find a complete line + // included in the touched line, this line is considered as + // the neighbor under certain conditions (see below) + + //--------------------------------------------------------------- + // + // + // __________________________________________________ 2Q + // | Q | + // | _________________________ |2P + // | |_____________P___________| => Second bottom |2P + 1 + // | line | + // |__________________________________________________|2Q + 1 + // => First encountered bottom line + // _________________________ ^ + // |_________________________| => Current line | Search direction + // | + //--------------------------------------------------------------- + + if (nbh_p) { - if (*nbh_p != literal::zero) // Not the background + std::vector<V> lines_nbh; + const V end_p = *nbh_p - 1; + const V* nbh_p_copy = nbh_p; + + for (; *nbh_p != end_p; ++nbh_p) { - if (*nbh_p % 2 == 1) // We have found the bottom of - // another line - lines_nbh.push_back(*nbh_p); - else + if (*nbh_p != literal::zero) // Not the background { - // We have found the top of a line. We are looking if - //we have already encountered the bottom of this - // line. If so, we link the current line with this one - // under certain conditions: - - if (std::find(lines_nbh.begin(), lines_nbh.end(), - *nbh_p + 1) != lines_nbh.end()) + if (*nbh_p % 2 == 1) // We have found the bottom of + // another line + lines_nbh.push_back(*nbh_p); + else { - // If we can link the complete line with the current line - if (// It must be in the search range - nbh_p < pstop - // Avoid loops - && right((*nbh_p >> 1) - 1) != i) - right(i) = (*nbh_p >> 1) - 1; - - // We have found a complete line, so we stop the search - break; + // We have found the top of a line. We are looking if + //we have already encountered the bottom of this + // line. If so, we link the current line with this one + // under certain conditions: + + if (std::find(lines_nbh.begin(), lines_nbh.end(), + *nbh_p + 1) != lines_nbh.end()) + { + // If we can link the complete line with the current line + if (// It must be in the search range + nbh_p < pstop + // Avoid loops + && right((*nbh_p >> 1) - 1) != l) + right(l) = (*nbh_p >> 1) - 1; + + // We have found a complete line, so we stop the search + break; + } } } } - } - // If we haven't found any included line in the first - // neighbor, then the line is considered as the neighbor of - // the current line + // If we haven't found any included line in the first + // neighbor, then the line is considered as the neighbor of + // the current line - if (*nbh_p == end_p) - right(i) = ((*nbh_p_copy - 1) >> 1) - 1; + if (*nbh_p == end_p) + right(l) = ((*nbh_p_copy - 1) >> 1) - 1; + } } } - } } //----------------------------------------------------------------------- @@ -950,76 +965,83 @@ namespace scribo template< typename L > inline - void finalize_links(util::array<value::int_u16>& left, - util::array<value::int_u16>& right, - const util::array< line_info<L> >& lines) + void finalize_links(line_links<L>& left, + line_links<L>& right, + const line_set<L>& lines) { - const unsigned nlines = lines.nelements(); - - for (unsigned i = 0; i < nlines; ++i) - { - const unsigned left_value = left(i); - const unsigned right_value = right(i); + // const unsigned nlines = lines.nelements(); - // If the right neighbor of my left neighbor is itself then its - // right neighbor is me + for_all_lines(l, lines) + if (lines(l).is_textline()) { - value::int_u16& v = right(left_value); + const unsigned left_value = left(l); + const unsigned right_value = right(l); - if (v == left_value) - v = i; - } + // If the right neighbor of my left neighbor is itself then its + // right neighbor is me + { + line_id_t& v = right(left_value); - // If the left neighbor of my right neighbor is itself then its - // left neighbor is me - { - value::int_u16& v = left(right_value); + if (v == left_value) + v = l; + } + + // If the left neighbor of my right neighbor is itself then its + // left neighbor is me + { + line_id_t& v = left(right_value); - if (v == right_value) - v = i; + if (v == right_value) + v = l; + } } - } } template <typename L> inline - void extract_paragraphs(line_set<L>& lines, - const image2d<bool>& input) + paragraph_set<L> + extract_paragraphs(line_set<L>& lines, + const image2d<bool>& input) { - typedef value::int_u16 V; + typedef scribo::def::lbl_type V; image2d<V> blocks(geom::rotate(input.domain(), -90, input.domain().pcenter())); data::fill(blocks, 0); - util::array< line_info<L> > lines_info; + // util::array< line_info<L> > lines_info; - for_all_lines(l, lines) - { - if (lines(l).is_textline()) - lines_info.append(lines(l)); - } + // for_all_lines(l, lines) + // { + // if (lines(l).is_textline()) + // lines_info.append(lines(l)); + // } - const unsigned nlines = lines_info.nelements(); +/// const unsigned nlines = lines_info.nelemnts(); util::array<box2d> rbbox; - util::array<V> left(nlines); - util::array<V> right(nlines); - util::array<V> output; + line_links<L> left(lines); + left(0) = 0; + line_links<L> right(lines); + right(0) = 0; + line_links<L> output(lines); + output(0) = 0; - rbbox.reserve(nlines); - output.reserve(nlines); + rbbox.resize(lines.nelements() + 1); std::cout << "Preparing lines" << std::endl; - prepare_lines(input.domain(), lines_info, blocks, rbbox); + prepare_lines(input.domain(), lines , blocks, rbbox); // io::pgm::save(blocks, "blocks.pgm"); std::cout << "Linking left" << std::endl; - process_left_link(blocks, rbbox, lines_info, left); + process_left_link(blocks, rbbox, lines , left); std::cout << "Linking right" << std::endl; - process_right_link(blocks, rbbox, lines_info, right); + process_right_link(blocks, rbbox, lines , right); std::cout << "Finalizing links" << std::endl; - finalize_links(left, right, lines_info); + finalize_links(left, right, lines ); // std::cout << "Finalizing merging" << std::endl; // finalize_line_merging(left, right, lines); std::cout << "Extracting paragraphs" << std::endl; - filter::paragraph_links(left, right, output, lines_info, input); + filter::paragraph_links(left, right, output, lines, input); + + paragraph_set<L> par_set = make::paragraph(output); + return par_set; } } diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index 25b328b..d60f3cc 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -60,6 +60,7 @@ # include <scribo/text/recognition.hh> # include <scribo/text/merging.hh> # include <scribo/text/link_lines.hh> +# include <scribo/text/paragraphs.hh> # include <scribo/make/debug_filename.hh> @@ -444,93 +445,96 @@ namespace scribo on_progress(); } - // Link text lines - on_new_progress_label("Linking text lines"); - line_links<L> llinks = scribo::text::link_lines(lines); - - - //===== DEBUG ===== -# ifndef SCRIBO_NDEBUG - if (debug::logger().is_enabled()) - { - 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); - } - - debug::logger().log_image(debug::AuxiliaryResults, - debug, "links_raw"); - } -# endif // ! SCRIBO_NDEBUG - //===== END OF DEBUG ===== - - on_progress(); - - - // Filter line links. - on_new_progress_label("Filter line links"); - llinks = scribo::filter::line_links_x_height(llinks); - - //===== DEBUG ===== -# ifndef SCRIBO_NDEBUG - if (debug::logger().is_enabled()) - { - 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); - - 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()); - - for_all_lines(i, lines) - { - if (! lines(i).is_textline()) - continue; - - mln::draw::box(debug, lines(i).bbox(), literal::red); - nbbox(llinks(i)).take(lines(i).bbox()); - } - - for (unsigned i = 1; i < nbbox.nelements(); ++i) - if (nbbox(i).is_valid()) - { - box2d b = nbbox(i).to_result(); - mln::draw::box(debug, b, literal::green); - b.enlarge(1); - mln::draw::box(debug, b, literal::green); - b.enlarge(1); - mln::draw::box(debug, b, literal::green); - } - - debug::logger().log_image(debug::AuxiliaryResults, - debug, "par"); - } -# endif // ! SCRIBO_NDEBUG - //===== END OF DEBUG ===== - - on_progress(); - - - // Construct paragraphs - on_new_progress_label("Constructing paragraphs"); - scribo::paragraph_set<L> parset = scribo::make::paragraph(llinks); +// // Link text lines +// on_new_progress_label("Linking text lines"); +// line_links<L> llinks = scribo::text::link_lines(lines); + + +// //===== DEBUG ===== +// # ifndef SCRIBO_NDEBUG +// if (debug::logger().is_enabled()) +// { +// 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); +// } + +// debug::logger().log_image(debug::AuxiliaryResults, +// debug, "links_raw"); +// } +// # endif // ! SCRIBO_NDEBUG +// //===== END OF DEBUG ===== + +// on_progress(); + + +// // Filter line links. +// on_new_progress_label("Filter line links"); +// llinks = scribo::filter::line_links_x_height(llinks); + +// //===== DEBUG ===== +// # ifndef SCRIBO_NDEBUG +// if (debug::logger().is_enabled()) +// { +// 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); + +// 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()); + +// for_all_lines(i, lines) +// { +// if (! lines(i).is_textline()) +// continue; + +// mln::draw::box(debug, lines(i).bbox(), literal::red); +// nbbox(llinks(i)).take(lines(i).bbox()); +// } + +// for (unsigned i = 1; i < nbbox.nelements(); ++i) +// if (nbbox(i).is_valid()) +// { +// box2d b = nbbox(i).to_result(); +// mln::draw::box(debug, b, literal::green); +// b.enlarge(1); +// mln::draw::box(debug, b, literal::green); +// b.enlarge(1); +// mln::draw::box(debug, b, literal::green); +// } + +// debug::logger().log_image(debug::AuxiliaryResults, +// debug, "par"); +// } +// # endif // ! SCRIBO_NDEBUG +// //===== END OF DEBUG ===== + +// on_progress(); + + +// // Construct paragraphs +// on_new_progress_label("Constructing paragraphs"); +// scribo::paragraph_set<L> parset = scribo::make::paragraph(llinks); + + scribo::paragraph_set<L> + parset = extract_paragraphs(lines, doc.binary_image()); doc.set_paragraphs(parset); on_progress(); -- 1.5.6.5

1 0

last-svn-commit-840-g2982cdf Correcting bug in stats
by Julien Marquegnies 04 May '11

04 May '11

--- scribo/scribo/core/stats.hh | 12 +++++++----- 1 files changed, 7 insertions(+), 5 deletions(-) diff --git a/scribo/scribo/core/stats.hh b/scribo/scribo/core/stats.hh index bc24044..88ecd68 100644 --- a/scribo/scribo/core/stats.hh +++ b/scribo/scribo/core/stats.hh @@ -261,7 +261,7 @@ private: unsigned i = 0; const unsigned nelements = data_.nelements(); - clusters[0] = cluster_index; + clusters.push_back(cluster_index); const T std = data_.standard_deviation(); for (i = 1; i < nelements - 1; ++i) @@ -276,11 +276,13 @@ private: clusters.push_back(cluster_index); } - if (nelements > 1 - && data_[i] - data_[i - 1] > std) - ++cluster_index; + if (nelements > 1) + { + if (data_[i] - data_[i - 1] > std) + ++cluster_index; - clusters.push_back(cluster_index); + clusters.push_back(cluster_index); + } clusters_.clear(); clusters_.reserve(cluster_index); -- 1.5.6.5

1 0