This patch update bioinspired opencl code to make it working
with latest opencv, major changes include:
1. data structure update, e.g. oclMat to UMat.
2. opencl kernel optimization to get it running faster
3. accuracy and perf test update.
Signed-off-by: Li Peng <peng.li@intel.com>
* Support getTextSize() to resolve issue #927
This getTextSize() funtion helps to print more text,
either on the same line or on another line.
To add getTextSize() function, some member variables are removed
which are corresponding at putText() function's arguments.
And a negative fontHeight must be asserted.
* Fix to remove whitespace
Bug in Text module when set Optimization Level to "-Ofast" in xcode (#921)
* Bug in Text module when set Optimization Level to "-Ofast" in xcode
Bug when set Optimization Level to "-Ofast" (Fastest, Aggressive Optimizations) in xcode, read the wrong patch_size parameters to 7 (should be 8). This bug will occured in loadOCRHMMClassifierCNN function called by example_text_character_recognition and example_text_segmented_word_recognition.
* Bug in Text module when set Optimization Level to "-Ofast" in xcode
Bug when set Optimization Level to "-Ofast" (Fastest, Aggressive Optimizations) in xcode, read the wrong patch_size parameters to 7 (should be 8). This bug will occured in loadOCRHMMClassifierCNN function called by example_text_character_recognition and example_text_segmented_word_recognition.
* Support freetype2
* Before call pkg_search_module() check it is available.
* If using libraries are unavailable, this module is shown in unavailable list.
* using ocv_define_modules() insted of target_link_libraries().
* Fix inner indentation in C++ namespaces
* Remove mImg from freetype2 and Create UserData class to refer in path drawing callbacks.
* to split interface and implementation details.
( and to avoid 3rd headers in public headers)
* PKG_CONFIG_FOUND is used instead of COMMAND okg_search_modules
* Update Document using Doxygen
* Add ocv_module_include_directories()
* Do no put pixels over image boundaries.
* Fix SIGSEGV (only valgrind) and missing text in top border line
* Fix trailing whitespace.
* Use Header File Macros
* Fix CMP0023 warning in cmake
* std::string to cv::String
* Fix trailing whitespace
* Change splitnumber from unsigned int to int to support python
* ocv_include_directories() should be define before ocv_define_module()
* Fix boundary problem in putTextBitmapBlend()
* Fix to draw an outline glyph which are on the boundary of image.
* Remove precomp.cpp
* Fix CMakeLists.txt to right order.
* Remove prototype definition of removed function.
* Add CV_Assert() where FreeType and Harfbuzz are called
* Add CV_Assert() to check InputOutputArray and line_type
This simple detector runs up to two times faster than LSD while achieving
competitive results. Please evaluate the algorithm with samples/fld_lines.cpp
Jeremy Nicola
Li Peng
Kumataro
Balint Cristian
arrybn
logic1988
StevenPuttemans
LaurentBerger
catree
Geoff
Suleyman TURKMEN
Vadim Pisarevsky
Alexander Alekhin
Jinhan Lee
Vladimir
Pavel Rojtberg
Pavel Vlasanek
Eugene Mamin
Vladislav Sovrasov
lukaszi89