[Olena-patches] last-svn-commit-535-g2eafecb Support for Tesseract 2.x and 3.x in Scribo.

* scribo/text/recognition.hh: Use the correct code depending on Tesseract version. * scribo/toolchain/nepomuk/text_extraction.hh: Add few comments about how to handle Tesseract versions. * src/Makefile.am, * src/text/Makefile.am, * src/toolchain/nepomuk/Makefile.am, * tests/toolchain/nepomuk/Makefile.am: Add path to config.h. * src/pbm_text_in_doc.cc, * src/text/pbm_lines_recognition.cc, * src/text/pbm_recognition.cc, * src/text_recognition_in_picture.cc, * src/toolchain/nepomuk/text_extraction.cc, * tests/toolchain/nepomuk/text_extraction.cc: Include config.h --- scribo/ChangeLog | 22 +++++++++ scribo/scribo/text/recognition.hh | 50 ++++++++++++++++++- scribo/scribo/toolchain/nepomuk/text_extraction.hh | 9 +++- scribo/src/Makefile.am | 6 ++- scribo/src/pbm_text_in_doc.cc | 5 ++- scribo/src/text/Makefile.am | 6 ++- scribo/src/text/pbm_lines_recognition.cc | 4 ++ scribo/src/text/pbm_recognition.cc | 4 ++ scribo/src/text_recognition_in_picture.cc | 4 ++ scribo/src/toolchain/nepomuk/Makefile.am | 3 +- scribo/src/toolchain/nepomuk/text_extraction.cc | 4 ++ scribo/tests/toolchain/nepomuk/Makefile.am | 3 +- scribo/tests/toolchain/nepomuk/text_extraction.cc | 4 ++ 13 files changed, 112 insertions(+), 12 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 82c02ca..cb05c44 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,27 @@ 2010-10-25 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; + Support for Tesseract 2.x and 3.x in Scribo. + + * scribo/text/recognition.hh: Use the correct code depending on + Tesseract version. + + * scribo/toolchain/nepomuk/text_extraction.hh: Add few comments + about how to handle Tesseract versions. + + * src/Makefile.am, + * src/text/Makefile.am, + * src/toolchain/nepomuk/Makefile.am, + * tests/toolchain/nepomuk/Makefile.am: Add path to config.h. + + * src/pbm_text_in_doc.cc, + * src/text/pbm_lines_recognition.cc, + * src/text/pbm_recognition.cc, + * src/text_recognition_in_picture.cc, + * src/toolchain/nepomuk/text_extraction.cc, + * tests/toolchain/nepomuk/text_extraction.cc: Include config.h + +2010-10-25 Guillaume Lazzara &lt;z(a)lrde.epita.fr&gt; + Cleanup files. * scribo/debug/usage.hh, diff --git a/scribo/scribo/text/recognition.hh b/scribo/scribo/text/recognition.hh index 17cbaa2..07f585d 100644 --- a/scribo/scribo/text/recognition.hh +++ b/scribo/scribo/text/recognition.hh @@ -61,6 +61,11 @@ # include <tesseract/baseapi.h> +# if !defined HAVE_TESSERACT_2 && !defined HAVE_TESSERACT_3 +# define HAVE_TESSERACT_2 +# endif + + namespace scribo { @@ -103,8 +108,19 @@ namespace scribo { trace::entering("scribo::text::recognition"); + // Initialize Tesseract. +# ifdef HAVE_TESSERACT_2 TessBaseAPI::InitWithLanguage(NULL, NULL, language, NULL, false, 0, NULL); +# else // HAVE_TESSERACT_3 + tesseract::TessBaseAPI tess; + if (tess.Init(NULL, language, NULL, 0, false) == -1) + { + std::cout << "Error: cannot initialize tesseract!" << std::endl; + abort(); + } + tess.SetPageSegMode(tesseract::PSM_SINGLE_LINE); +# endif // HAVE_TESSERACT_2 typedef mln_ch_value(L,bool) I; int vals[] = { 0, 9, 0, 9, 0, @@ -155,8 +171,8 @@ namespace scribo data::fill(line_image, false); data::paste_without_localization(text_ima, line_image); - // Recognize characters. +# ifdef HAVE_TESSERACT_2 char* s = TessBaseAPI::TesseractRect( (unsigned char*) line_image.buffer(), sizeof (bool), // Pixel size. @@ -165,7 +181,16 @@ namespace scribo 0, // Top line_image.ncols(), // n cols line_image.nrows()); // n rows - +# else // HAVE_TESSERACT_3 + char* s = tess.TesseractRect( + (unsigned char*) line_image.buffer(), + sizeof (bool), // Pixel size. + line_image.ncols() * sizeof (bool), // Row_offset + 0, // Left + 0, // Top + line_image.ncols(), // n cols + line_image.nrows()); // n rows +# endif // ! HAVE_TESSERACT_2 if (s != 0) { @@ -194,7 +219,16 @@ namespace scribo mln_precondition(line.is_valid()); // Initialize Tesseract. +# ifdef HAVE_TESSERACT_2 TessBaseAPI::InitWithLanguage(NULL, NULL, language, NULL, false, 0, NULL); +# else // HAVE_TESSERACT_3 + tesseract::TessBaseAPI tess; + if (tess.Init(NULL, language, NULL, 0, false) == -1) + { + std::cout << "Error: cannot initialize tesseract!" << std::endl; + abort(); + } +# endif // ! HAVE_TESSERACT_2 std::ofstream file; if (!output_file.empty()) @@ -213,6 +247,7 @@ namespace scribo border::resize(text_ima, 0); // Recognize characters. +# ifdef HAVE_TESSERACT_2 char* s = TessBaseAPI::TesseractRect( (unsigned char*) text_ima.buffer(), sizeof (bool), // Pixel size. @@ -221,7 +256,16 @@ namespace scribo 0, // Top text_ima.ncols(), // n cols text_ima.nrows()); // n rows - +# else // HAVE_TESSERACT_3 + char* s = tess.TesseractRect( + (unsigned char*) text_ima.buffer(), + sizeof (bool), // Pixel size. + text_ima.ncols() * sizeof (bool), // Row_offset + 0, // Left + 0, // Top + text_ima.ncols(), // n cols + text_ima.nrows()); // n rows +# endif // ! HAVE_TESSERACT_2 if (s != 0) { diff --git a/scribo/scribo/toolchain/nepomuk/text_extraction.hh b/scribo/scribo/toolchain/nepomuk/text_extraction.hh index e16cd56..effb13f 100644 --- a/scribo/scribo/toolchain/nepomuk/text_extraction.hh +++ b/scribo/scribo/toolchain/nepomuk/text_extraction.hh @@ -64,12 +64,17 @@ namespace scribo This is a convenient routine to be used in Nepomuk. - - \param[in] ima A document image. The \return A set of recognized words. + + Don't forget to define NDEBUG for compilation to disable debug + checks. + + Depending on your version of Tesseract (OCR) you may define + HAVE_TESSERACT_2 or HAVE_TESSERACT_3 . + */ QSet<QString> text_extraction(const QImage& input); diff --git a/scribo/src/Makefile.am b/scribo/src/Makefile.am index d3f686b..c99a791 100644 --- a/scribo/src/Makefile.am +++ b/scribo/src/Makefile.am @@ -42,7 +42,8 @@ if HAVE_TESSERACT pbm_text_in_doc_SOURCES = pbm_text_in_doc.cc pbm_text_in_doc_CPPFLAGS = $(AM_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ - $(TIFF_CPPFLAGS) + $(TIFF_CPPFLAGS) \ + -I$(top_builddir) pbm_text_in_doc_LDFLAGS = $(AM_LDFLAGS) \ $(TESSERACT_LDFLAGS) \ $(TIFF_LDFLAGS) \ @@ -77,7 +78,8 @@ if HAVE_TESSERACT text_recognition_in_picture_CPPFLAGS = $(AM_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ $(TIFF_CPPFLAGS) \ - $(MAGICKXX_CPPFLAGS) + $(MAGICKXX_CPPFLAGS) \ + -I$(top_builddir) text_recognition_in_picture_LDFLAGS = $(AM_LDFLAGS) \ -lpthread \ $(TESSERACT_LDFLAGS) \ diff --git a/scribo/src/pbm_text_in_doc.cc b/scribo/src/pbm_text_in_doc.cc index fa8b5ae..23ed9e7 100644 --- a/scribo/src/pbm_text_in_doc.cc +++ b/scribo/src/pbm_text_in_doc.cc @@ -24,6 +24,9 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif #include <libgen.h> #include <fstream> @@ -119,7 +122,7 @@ int main(int argc, char* argv[]) typedef image2d<scribo::def::lbl_type> L; line_set<L> lines = scribo::toolchain::text_in_doc(input, denoise, debug); - + // Saving results scribo::io::xml::save(argv[1], lines, "out.xml", true); diff --git a/scribo/src/text/Makefile.am b/scribo/src/text/Makefile.am index fc03ae8..dbdb1da 100644 --- a/scribo/src/text/Makefile.am +++ b/scribo/src/text/Makefile.am @@ -30,7 +30,8 @@ if HAVE_TIFF pbm_lines_recognition_SOURCES = pbm_lines_recognition.cc pbm_lines_recognition_CPPFLAGS = $(AM_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ - $(TIFF_CPPFLAGS) + $(TIFF_CPPFLAGS) \ + -I$(top_builddir) pbm_lines_recognition_LDFLAGS = $(AM_LDFLAGS) \ $(TESSERACT_LDFLAGS) \ $(TIFF_LDFLAGS) \ @@ -40,7 +41,8 @@ if HAVE_TIFF pbm_recognition_SOURCES = pbm_recognition.cc pbm_recognition_CPPFLAGS = $(AM_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ - $(TIFF_CPPFLAGS) + $(TIFF_CPPFLAGS) \ + -I$(top_builddir) pbm_recognition_LDFLAGS = $(AM_LDFLAGS) \ $(TESSERACT_LDFLAGS) \ $(TIFF_LDFLAGS) \ diff --git a/scribo/src/text/pbm_lines_recognition.cc b/scribo/src/text/pbm_lines_recognition.cc index 5aaa15b..3451449 100644 --- a/scribo/src/text/pbm_lines_recognition.cc +++ b/scribo/src/text/pbm_lines_recognition.cc @@ -23,6 +23,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + #include <iostream> #include <mln/core/image/image2d.hh> diff --git a/scribo/src/text/pbm_recognition.cc b/scribo/src/text/pbm_recognition.cc index 082faef..b262aab 100644 --- a/scribo/src/text/pbm_recognition.cc +++ b/scribo/src/text/pbm_recognition.cc @@ -23,6 +23,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + #include <iostream> #include <mln/core/image/image2d.hh> diff --git a/scribo/src/text_recognition_in_picture.cc b/scribo/src/text_recognition_in_picture.cc index acf8fec..b94ed0b 100644 --- a/scribo/src/text_recognition_in_picture.cc +++ b/scribo/src/text_recognition_in_picture.cc @@ -24,6 +24,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + #include <libgen.h> #include <iostream> diff --git a/scribo/src/toolchain/nepomuk/Makefile.am b/scribo/src/toolchain/nepomuk/Makefile.am index 07c2c33..4328be5 100644 --- a/scribo/src/toolchain/nepomuk/Makefile.am +++ b/scribo/src/toolchain/nepomuk/Makefile.am @@ -30,7 +30,8 @@ text_extraction_CPPFLAGS = $(AM_CPPFLAGS) \ $(MAGICKXX_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ $(TIFF_CPPFLAGS) \ - $(QT_CPPFLAGS) + $(QT_CPPFLAGS) \ + -I$(top_builddir) text_extraction_LDFLAGS = $(AM_LDFLAGS) \ $(MAGICKXX_LDFLAGS) \ $(TESSERACT_LDFLAGS) \ diff --git a/scribo/src/toolchain/nepomuk/text_extraction.cc b/scribo/src/toolchain/nepomuk/text_extraction.cc index e5e2de4..41e843a 100644 --- a/scribo/src/toolchain/nepomuk/text_extraction.cc +++ b/scribo/src/toolchain/nepomuk/text_extraction.cc @@ -23,6 +23,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + #include <libgen.h> #include <iostream> #include <QtCore/QDebug> diff --git a/scribo/tests/toolchain/nepomuk/Makefile.am b/scribo/tests/toolchain/nepomuk/Makefile.am index 4bce3bd..67a594b 100644 --- a/scribo/tests/toolchain/nepomuk/Makefile.am +++ b/scribo/tests/toolchain/nepomuk/Makefile.am @@ -31,7 +31,8 @@ text_extraction_SOURCES = text_extraction.cc text_extraction_CXXFLAGS = $(QT_CXXFLAGS) $(AM_CXXFLAGS) text_extraction_CPPFLAGS = $(QT_CPPFLAGS) $(AM_CPPFLAGS) \ $(TESSERACT_CPPFLAGS) \ - $(TIFF_CPPFLAGS) + $(TIFF_CPPFLAGS) \ + -I$(top_builddir) text_extraction_LDFLAGS = $(QT_LDFLAGS) $(LDFLAGS) \ $(TESSERACT_LDFLAGS) \ $(TIFF_LDFLAGS) \ diff --git a/scribo/tests/toolchain/nepomuk/text_extraction.cc b/scribo/tests/toolchain/nepomuk/text_extraction.cc index 027ad99..c475aa9 100644 --- a/scribo/tests/toolchain/nepomuk/text_extraction.cc +++ b/scribo/tests/toolchain/nepomuk/text_extraction.cc @@ -27,6 +27,10 @@ /// /// Test of scribo::toolchain::nepomuk::text_extraction +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + #include <QtGui/QImage> #include <QtCore> #include <scribo/toolchain/nepomuk/text_extraction.hh> -- 1.5.6.5