Merge pull request #1761 from bhuroc:qtkit-video-file-2.4

diff --git a/.gitignore b/.gitignore
index 94537d6..46d3499 100644 (file)
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,5 @@
 .*.swp
 .DS_Store
 .sw[a-z]
-/modules/refman.rst
 tags
 tegra/

diff --git a/3rdparty/ffmpeg/make.bat b/3rdparty/ffmpeg/make.bat
index 442fce5..318c2fe 100644 (file)
--- a/3rdparty/ffmpeg/make.bat
+++ b/3rdparty/ffmpeg/make.bat
@@ -1,2 +1,2 @@
-set path=c:\dev\msys32\bin;%path% & gcc -Wall -shared -o opencv_ffmpeg.dll -O2 -x c++ -I../include -I../include/ffmpeg_ -I../../modules/highgui/src ffopencv.c -L../lib -lavformat -lavcodec -lavdevice -lswscale -lavutil -liconv -lws2_32
+set path=c:\dev\msys32\bin;%path% & gcc -Wall -shared -o opencv_ffmpeg.dll -O2 -x c++ -I../include -I../include/ffmpeg_ -I../../modules/highgui/src ffopencv.c -L../lib -lavformat -lavcodec -lavdevice -lswscale -lavutil -lws2_32
 set path=c:\dev\msys64\bin;%path% & gcc -m64 -Wall -shared -o opencv_ffmpeg_64.dll -O2 -x c++ -I../include -I../include/ffmpeg_ -I../../modules/highgui/src ffopencv.c -L../lib -lavformat64 -lavcodec64 -lavdevice64 -lswscale64 -lavutil64 -lws2_32
\ No newline at end of file

diff --git a/3rdparty/ffmpeg/opencv_ffmpeg.dll b/3rdparty/ffmpeg/opencv_ffmpeg.dll
index d556b0d..b1e70df 100644 (file)
Binary files a/3rdparty/ffmpeg/opencv_ffmpeg.dll and b/3rdparty/ffmpeg/opencv_ffmpeg.dll differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r2.2.0.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r2.2.0.so
index 6f28f2c..faff854 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r2.2.0.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r2.2.0.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r2.3.3.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r2.3.3.so
index 010641e..b2dfbad 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r2.3.3.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r2.3.3.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r3.0.1.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r3.0.1.so
index 5a145b2..b556709 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r3.0.1.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r3.0.1.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.0.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.0.so
index a524b74..b4636df 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.0.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.0.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.3.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.3.so
index a1802f1..9068d3b 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.3.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.0.3.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.1.1.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.1.1.so
index 089c7e9..140c9d5 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.1.1.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.1.1.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.2.0.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.2.0.so
index a9ffa4b..ca5cbb9 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.2.0.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.2.0.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.3.0.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.3.0.so
index 8ff7177..433a3b9 100755 (executable)
Binary files a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.3.0.so and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.3.0.so differ

diff --git a/3rdparty/lib/armeabi-v7a/libnative_camera_r4.4.0.so b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.4.0.so
new file mode 100755 (executable)
index 0000000..f6fb79a
Binary files /dev/null and b/3rdparty/lib/armeabi-v7a/libnative_camera_r4.4.0.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r2.2.0.so b/3rdparty/lib/armeabi/libnative_camera_r2.2.0.so
index b6ce0d5..e00ff4a 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r2.2.0.so and b/3rdparty/lib/armeabi/libnative_camera_r2.2.0.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r2.3.3.so b/3rdparty/lib/armeabi/libnative_camera_r2.3.3.so
index 635ce68..5e6fa00 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r2.3.3.so and b/3rdparty/lib/armeabi/libnative_camera_r2.3.3.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r3.0.1.so b/3rdparty/lib/armeabi/libnative_camera_r3.0.1.so
index caacf39..1c095e1 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r3.0.1.so and b/3rdparty/lib/armeabi/libnative_camera_r3.0.1.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.0.0.so b/3rdparty/lib/armeabi/libnative_camera_r4.0.0.so
index fff4a80..5e543a9 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r4.0.0.so and b/3rdparty/lib/armeabi/libnative_camera_r4.0.0.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.0.3.so b/3rdparty/lib/armeabi/libnative_camera_r4.0.3.so
index 3119265..a89085a 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r4.0.3.so and b/3rdparty/lib/armeabi/libnative_camera_r4.0.3.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.1.1.so b/3rdparty/lib/armeabi/libnative_camera_r4.1.1.so
index 7c18baf..ba479d3 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r4.1.1.so and b/3rdparty/lib/armeabi/libnative_camera_r4.1.1.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.2.0.so b/3rdparty/lib/armeabi/libnative_camera_r4.2.0.so
index 8bb093a..bb25fe0 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r4.2.0.so and b/3rdparty/lib/armeabi/libnative_camera_r4.2.0.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.3.0.so b/3rdparty/lib/armeabi/libnative_camera_r4.3.0.so
index a05f179..53f5092 100755 (executable)
Binary files a/3rdparty/lib/armeabi/libnative_camera_r4.3.0.so and b/3rdparty/lib/armeabi/libnative_camera_r4.3.0.so differ

diff --git a/3rdparty/lib/armeabi/libnative_camera_r4.4.0.so b/3rdparty/lib/armeabi/libnative_camera_r4.4.0.so
new file mode 100755 (executable)
index 0000000..08a9cfb
Binary files /dev/null and b/3rdparty/lib/armeabi/libnative_camera_r4.4.0.so differ

diff --git a/3rdparty/lib/libavcodec.a b/3rdparty/lib/libavcodec.a
index 2f8bace..45c31c5 100644 (file)
Binary files a/3rdparty/lib/libavcodec.a and b/3rdparty/lib/libavcodec.a differ

diff --git a/3rdparty/lib/libavdevice.a b/3rdparty/lib/libavdevice.a
index 15ef35a..41ba6a1 100644 (file)
Binary files a/3rdparty/lib/libavdevice.a and b/3rdparty/lib/libavdevice.a differ

diff --git a/3rdparty/lib/libavformat.a b/3rdparty/lib/libavformat.a
index 9cd2715..ab267be 100644 (file)
Binary files a/3rdparty/lib/libavformat.a and b/3rdparty/lib/libavformat.a differ

diff --git a/3rdparty/lib/libavutil.a b/3rdparty/lib/libavutil.a
index cfb89bd..7c0cda3 100644 (file)
Binary files a/3rdparty/lib/libavutil.a and b/3rdparty/lib/libavutil.a differ

diff --git a/3rdparty/lib/libswscale.a b/3rdparty/lib/libswscale.a
index 23e6d8a..91f9d75 100644 (file)
Binary files a/3rdparty/lib/libswscale.a and b/3rdparty/lib/libswscale.a differ

diff --git a/3rdparty/lib/mips/libnative_camera_r4.0.3.so b/3rdparty/lib/mips/libnative_camera_r4.0.3.so
index b950044..9c467b3 100755 (executable)
Binary files a/3rdparty/lib/mips/libnative_camera_r4.0.3.so and b/3rdparty/lib/mips/libnative_camera_r4.0.3.so differ

diff --git a/3rdparty/lib/mips/libnative_camera_r4.1.1.so b/3rdparty/lib/mips/libnative_camera_r4.1.1.so
index d11dcf0..65e6fc6 100755 (executable)
Binary files a/3rdparty/lib/mips/libnative_camera_r4.1.1.so and b/3rdparty/lib/mips/libnative_camera_r4.1.1.so differ

diff --git a/3rdparty/lib/mips/libnative_camera_r4.2.0.so b/3rdparty/lib/mips/libnative_camera_r4.2.0.so
index b06a681..7970810 100755 (executable)
Binary files a/3rdparty/lib/mips/libnative_camera_r4.2.0.so and b/3rdparty/lib/mips/libnative_camera_r4.2.0.so differ

diff --git a/3rdparty/lib/mips/libnative_camera_r4.3.0.so b/3rdparty/lib/mips/libnative_camera_r4.3.0.so
index 844b806..38e0f18 100755 (executable)
Binary files a/3rdparty/lib/mips/libnative_camera_r4.3.0.so and b/3rdparty/lib/mips/libnative_camera_r4.3.0.so differ

diff --git a/3rdparty/lib/mips/libnative_camera_r4.4.0.so b/3rdparty/lib/mips/libnative_camera_r4.4.0.so
new file mode 100755 (executable)
index 0000000..4545398
Binary files /dev/null and b/3rdparty/lib/mips/libnative_camera_r4.4.0.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r2.3.3.so b/3rdparty/lib/x86/libnative_camera_r2.3.3.so
index 0dd8904..e24f344 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r2.3.3.so and b/3rdparty/lib/x86/libnative_camera_r2.3.3.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r3.0.1.so b/3rdparty/lib/x86/libnative_camera_r3.0.1.so
index 105a19d..6dbeaea 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r3.0.1.so and b/3rdparty/lib/x86/libnative_camera_r3.0.1.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r4.0.3.so b/3rdparty/lib/x86/libnative_camera_r4.0.3.so
index b01a4bd..f789896 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r4.0.3.so and b/3rdparty/lib/x86/libnative_camera_r4.0.3.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r4.1.1.so b/3rdparty/lib/x86/libnative_camera_r4.1.1.so
index a59ae39..0039050 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r4.1.1.so and b/3rdparty/lib/x86/libnative_camera_r4.1.1.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r4.2.0.so b/3rdparty/lib/x86/libnative_camera_r4.2.0.so
index b90b826..a3d96ba 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r4.2.0.so and b/3rdparty/lib/x86/libnative_camera_r4.2.0.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r4.3.0.so b/3rdparty/lib/x86/libnative_camera_r4.3.0.so
index 6607e5d..2770f04 100755 (executable)
Binary files a/3rdparty/lib/x86/libnative_camera_r4.3.0.so and b/3rdparty/lib/x86/libnative_camera_r4.3.0.so differ

diff --git a/3rdparty/lib/x86/libnative_camera_r4.4.0.so b/3rdparty/lib/x86/libnative_camera_r4.4.0.so
new file mode 100755 (executable)
index 0000000..83a6016
Binary files /dev/null and b/3rdparty/lib/x86/libnative_camera_r4.4.0.so differ

diff --git a/cmake/OpenCVGenConfig.cmake b/cmake/OpenCVGenConfig.cmake
index 362841b..411d225 100644 (file)
--- a/cmake/OpenCVGenConfig.cmake
+++ b/cmake/OpenCVGenConfig.cmake
@@ -134,11 +134,11 @@ if(WIN32)
   configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/OpenCVConfig.cmake.in" "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig.cmake" IMMEDIATE @ONLY)
   configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/OpenCVConfig-version.cmake.in" "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig-version.cmake" IMMEDIATE @ONLY)
   if(BUILD_SHARED_LIBS)
-    install(FILES "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig.cmake" DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}/lib")
-    install(EXPORT OpenCVModules DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}/lib" FILE OpenCVModules${modules_file_suffix}.cmake)
+    install(FILES "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig.cmake" DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}lib")
+    install(EXPORT OpenCVModules DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}lib" FILE OpenCVModules${modules_file_suffix}.cmake)
   else()
-    install(FILES "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig.cmake" DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}/staticlib")
-    install(EXPORT OpenCVModules DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}/staticlib" FILE OpenCVModules${modules_file_suffix}.cmake)
+    install(FILES "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig.cmake" DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}staticlib")
+    install(EXPORT OpenCVModules DESTINATION "${OpenCV_INSTALL_BINARIES_PREFIX}staticlib" FILE OpenCVModules${modules_file_suffix}.cmake)
   endif()
   install(FILES "${CMAKE_BINARY_DIR}/win-install/OpenCVConfig-version.cmake" DESTINATION "${CMAKE_INSTALL_PREFIX}")
   install(FILES "${OpenCV_SOURCE_DIR}/cmake/OpenCVConfig.cmake" DESTINATION "${CMAKE_INSTALL_PREFIX}/")

diff --git a/cmake/OpenCVGenHeaders.cmake b/cmake/OpenCVGenHeaders.cmake
index c7129fe..35da0fb 100644 (file)
--- a/cmake/OpenCVGenHeaders.cmake
+++ b/cmake/OpenCVGenHeaders.cmake
@@ -1,5 +1,5 @@
 # platform-specific config file
-configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/cvconfig.h.cmake" "${OPENCV_CONFIG_FILE_INCLUDE_DIR}/cvconfig.h")
+configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/cvconfig.h.in" "${OPENCV_CONFIG_FILE_INCLUDE_DIR}/cvconfig.h")
 
 # ----------------------------------------------------------------------------
 #  opencv_modules.hpp based on actual modules list

diff --git a/cmake/OpenCVGenPkgconfig.cmake b/cmake/OpenCVGenPkgconfig.cmake
index a36b70e..cd54f11 100644 (file)
--- a/cmake/OpenCVGenPkgconfig.cmake
+++ b/cmake/OpenCVGenPkgconfig.cmake
@@ -76,7 +76,9 @@ if(INSTALL_TO_MANGLED_PATHS)
 else()
   set(OPENCV_PC_FILE_NAME opencv.pc)
 endif()
-configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/opencv-XXX.pc.cmake.in" "${CMAKE_BINARY_DIR}/unix-install/${OPENCV_PC_FILE_NAME}" @ONLY IMMEDIATE)
+configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/opencv-XXX.pc.in"
+               "${CMAKE_BINARY_DIR}/unix-install/${OPENCV_PC_FILE_NAME}"
+               @ONLY IMMEDIATE)
 
 if(UNIX AND NOT ANDROID)
   install(FILES ${CMAKE_BINARY_DIR}/unix-install/${OPENCV_PC_FILE_NAME} DESTINATION ${OPENCV_LIB_INSTALL_PATH}/pkgconfig)

diff --git a/cmake/templates/OpenCVConfig.cmake.in b/cmake/templates/OpenCVConfig.cmake.in
index ee1eb59..91791b6 100644 (file)
--- a/cmake/templates/OpenCVConfig.cmake.in
+++ b/cmake/templates/OpenCVConfig.cmake.in
@@ -41,7 +41,9 @@ if(ANDROID)
   string(REPLACE - _ modules_file_suffix "_${ANDROID_NDK_ABI_NAME}")
 endif()
 
-include(${CMAKE_CURRENT_LIST_DIR}/OpenCVModules${modules_file_suffix}.cmake)
+if(NOT TARGET opencv_core)
+  include(${CMAKE_CURRENT_LIST_DIR}/OpenCVModules${modules_file_suffix}.cmake)
+endif()
 
 # TODO All things below should be reviewed. What is about of moving this code into related modules (special vars/hooks/files)
 

diff --git a/cmake/templates/cvconfig.h.cmake b/cmake/templates/cvconfig.h.in
similarity index 100%
rename from cmake/templates/cvconfig.h.cmake
rename to cmake/templates/cvconfig.h.in

diff --git a/cmake/templates/opencv-XXX.pc.cmake.in b/cmake/templates/opencv-XXX.pc.in
similarity index 100%
rename from cmake/templates/opencv-XXX.pc.cmake.in
rename to cmake/templates/opencv-XXX.pc.in

diff --git a/doc/CMakeLists.txt b/doc/CMakeLists.txt
index 45dd7e4..5793f72 100644 (file)
--- a/doc/CMakeLists.txt
+++ b/doc/CMakeLists.txt
@@ -2,9 +2,6 @@
 #  CMake file for OpenCV docs
 #
 
-file(GLOB FILES_TEX *.tex *.sty *.bib)
-file(GLOB FILES_TEX_PICS pics/*.png pics/*.jpg)
-
 if(BUILD_DOCS AND HAVE_SPHINX)
 
   project(opencv_docs)
@@ -23,59 +20,84 @@ if(BUILD_DOCS AND HAVE_SPHINX)
   set(OPTIONAL_DOC_LIST "")
 
 
-  set(OPENCV2_BASE_MODULES core imgproc highgui video calib3d features2d objdetect ml flann gpu photo stitching nonfree contrib legacy)
-
   # build lists of modules to be documented
-  set(OPENCV2_MODULES "")
-  set(OPENCV_MODULES "")
+  set(BASE_MODULES "")
+  set(EXTRA_MODULES "")
 
   foreach(mod ${OPENCV_MODULES_BUILD} ${OPENCV_MODULES_DISABLED_USER} ${OPENCV_MODULES_DISABLED_AUTO} ${OPENCV_MODULES_DISABLED_FORCE})
     string(REGEX REPLACE "^opencv_" "" mod "${mod}")
     if("${OPENCV_MODULE_opencv_${mod}_LOCATION}" STREQUAL "${OpenCV_SOURCE_DIR}/modules/${mod}")
-      list(APPEND OPENCV2_MODULES ${mod})
+      list(APPEND BASE_MODULES ${mod})
     else()
-      list(APPEND OPENCV_MODULES ${mod})
+      list(APPEND EXTRA_MODULES ${mod})
     endif()
   endforeach()
-  list(REMOVE_ITEM OPENCV2_MODULES ${OPENCV2_BASE_MODULES})
-  ocv_list_sort(OPENCV2_MODULES)
-  ocv_list_sort(OPENCV_MODULES)
+
+  set(FIXED_ORDER_MODULES core imgproc highgui video calib3d features2d objdetect ml flann gpu photo stitching nonfree contrib legacy)
+
+  list(REMOVE_ITEM BASE_MODULES ${FIXED_ORDER_MODULES})
+
+  ocv_list_sort(BASE_MODULES)
+  ocv_list_sort(EXTRA_MODULES)
+
+  set(BASE_MODULES ${FIXED_ORDER_MODULES} ${BASE_MODULES})
 
   # build lists of documentation files and generate table of contents for reference manual
-  set(OPENCV_FILES_REF "")
-  set(OPENCV_FILES_REF_PICT "")
-  set(OPENCV_REFMAN_TOC "")
 
-  foreach(mod ${OPENCV2_BASE_MODULES} ${OPENCV2_MODULES} ${OPENCV_MODULES})
-    file(GLOB_RECURSE _OPENCV_FILES_REF "${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc/*.rst")
-    file(GLOB_RECURSE _OPENCV_FILES_REF_PICT "${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc/*.png" "${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc/*.jpg")
-    list(APPEND OPENCV_FILES_REF ${_OPENCV_FILES_REF})
-    list(APPEND OPENCV_FILES_REF_PICT ${_OPENCV_FILES_REF_PICT})
+  set(DOC_FAKE_ROOT "${CMAKE_CURRENT_BINARY_DIR}/fake-root")
+  set(DOC_FAKE_ROOT_FILES "")
 
-    set(toc_file "${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc/${mod}.rst")
-    if(EXISTS "${toc_file}")
-      file(RELATIVE_PATH toc_file "${OpenCV_SOURCE_DIR}/modules" "${toc_file}")
-      set(OPENCV_REFMAN_TOC "${OPENCV_REFMAN_TOC}   ${toc_file}\n")
+  function(ocv_doc_add_file_to_fake_root source destination)
+    add_custom_command(
+      OUTPUT "${DOC_FAKE_ROOT}/${destination}"
+      COMMAND "${CMAKE_COMMAND}" -E copy "${source}" "${DOC_FAKE_ROOT}/${destination}"
+      DEPENDS "${source}"
+      COMMENT "Copying ${destination} to fake root..."
+      VERBATIM
+    )
+    list(APPEND DOC_FAKE_ROOT_FILES "${DOC_FAKE_ROOT}/${destination}")
+    set(DOC_FAKE_ROOT_FILES "${DOC_FAKE_ROOT_FILES}" PARENT_SCOPE)
+  endfunction()
+
+  function(ocv_doc_add_to_fake_root source)
+    if(ARGC GREATER 1)
+      set(destination "${ARGV1}")
+    else()
+      file(RELATIVE_PATH destination "${OpenCV_SOURCE_DIR}" "${source}")
+    endif()
+
+    if(IS_DIRECTORY "${source}")
+      file(GLOB_RECURSE files RELATIVE "${source}" "${source}/*")
+
+      foreach(file ${files})
+        ocv_doc_add_file_to_fake_root("${source}/${file}" "${destination}/${file}")
+      endforeach()
+    else()
+      ocv_doc_add_file_to_fake_root("${source}" "${destination}")
     endif()
-  endforeach()
 
-  file(GLOB_RECURSE _OPENCV_FILES_REF "${OpenCV_SOURCE_DIR}/platforms/android/service/doc/*.rst")
-  file(GLOB_RECURSE _OPENCV_FILES_REF_PICT "${OpenCV_SOURCE_DIR}/platforms/android/service/doc/*.png" "${OpenCV_SOURCE_DIR}/platforms/android/service/doc/*.jpg")
-  list(APPEND OPENCV_FILES_REF ${_OPENCV_FILES_REF})
-  list(APPEND OPENCV_FILES_REF_PICT ${_OPENCV_FILES_REF_PICT})
+    set(DOC_FAKE_ROOT_FILES "${DOC_FAKE_ROOT_FILES}" PARENT_SCOPE)
+  endfunction()
 
-  configure_file("${OpenCV_SOURCE_DIR}/modules/refman.rst.in" "${OpenCV_SOURCE_DIR}/modules/refman.rst" IMMEDIATE @ONLY)
+  set(OPENCV_REFMAN_TOC "")
+
+  foreach(mod ${BASE_MODULES} ${EXTRA_MODULES})
+    if(EXISTS "${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc/${mod}.rst")
+      ocv_doc_add_to_fake_root("${OPENCV_MODULE_opencv_${mod}_LOCATION}/doc" modules/${mod}/doc)
+      set(OPENCV_REFMAN_TOC "${OPENCV_REFMAN_TOC}   ${mod}/doc/${mod}.rst\n")
+    endif()
+  endforeach()
 
-  file(GLOB_RECURSE OPENCV_FILES_UG  user_guide/*.rst)
-  file(GLOB_RECURSE OPENCV_FILES_TUT tutorials/*.rst)
-  file(GLOB_RECURSE OPENCV_FILES_TUT_PICT tutorials/*.png tutorials/*.jpg)
+  configure_file("${OpenCV_SOURCE_DIR}/modules/refman.rst.in" "${DOC_FAKE_ROOT}/modules/refman.rst" @ONLY)
 
-  set(OPENCV_DOC_DEPS conf.py ${OPENCV_FILES_REF} ${OPENCV_FILES_REF_PICT}
-           ${OPENCV_FILES_UG} ${OPENCV_FILES_TUT} ${OPENCV_FILES_TUT_PICT})
+  ocv_doc_add_to_fake_root("${OpenCV_SOURCE_DIR}/index.rst")
+  ocv_doc_add_to_fake_root("${OpenCV_SOURCE_DIR}/doc")
+  ocv_doc_add_to_fake_root("${OpenCV_SOURCE_DIR}/platforms/android")
+  ocv_doc_add_to_fake_root("${OpenCV_SOURCE_DIR}/samples")
 
   if(PDFLATEX_COMPILER)
     add_custom_target(docs
-      COMMAND ${SPHINX_BUILD} -b latex -c ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/.. .
+      COMMAND ${SPHINX_BUILD} -b latex -c "${CMAKE_CURRENT_SOURCE_DIR}" "${DOC_FAKE_ROOT}" .
       COMMAND ${CMAKE_COMMAND} -E copy_directory ${CMAKE_CURRENT_SOURCE_DIR}/pics ${CMAKE_CURRENT_BINARY_DIR}/doc/opencv1/pics
       COMMAND ${CMAKE_COMMAND} -E copy_if_different ${CMAKE_CURRENT_SOURCE_DIR}/mymath.sty ${CMAKE_CURRENT_BINARY_DIR}
       COMMAND ${PYTHON_EXECUTABLE} "${CMAKE_CURRENT_SOURCE_DIR}/patch_refman_latex.py" opencv2refman.tex
@@ -95,7 +117,7 @@ if(BUILD_DOCS AND HAVE_SPHINX)
       COMMAND ${CMAKE_COMMAND} -E echo "Generating opencv_cheatsheet.pdf"
       COMMAND ${PDFLATEX_COMPILER} -interaction=batchmode "${CMAKE_CURRENT_SOURCE_DIR}/opencv_cheatsheet.tex"
       COMMAND ${PDFLATEX_COMPILER} -interaction=batchmode "${CMAKE_CURRENT_SOURCE_DIR}/opencv_cheatsheet.tex"
-      DEPENDS ${OPENCV_DOC_DEPS}
+      DEPENDS ${DOC_FAKE_ROOT_FILES}
       WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
       COMMENT "Generating the PDF Manuals"
       )
@@ -109,9 +131,9 @@ if(BUILD_DOCS AND HAVE_SPHINX)
   endif()
 
   add_custom_target(html_docs
-    COMMAND ${SPHINX_BUILD} -b html -c ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/.. ./_html
+    COMMAND "${SPHINX_BUILD}" -b html -c "${CMAKE_CURRENT_SOURCE_DIR}" "${DOC_FAKE_ROOT}" ./_html
     COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/mymath.sty ${CMAKE_CURRENT_BINARY_DIR}
-    DEPENDS ${OPENCV_DOC_DEPS}
+    DEPENDS ${DOC_FAKE_ROOT_FILES}
     WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
     COMMENT "Generating Online Documentation"
     )

diff --git a/modules/androidcamera/camera_wrapper/camera_wrapper.cpp b/modules/androidcamera/camera_wrapper/camera_wrapper.cpp
index ca631fc..03796d1 100644 (file)
--- a/modules/androidcamera/camera_wrapper/camera_wrapper.cpp
+++ b/modules/androidcamera/camera_wrapper/camera_wrapper.cpp
@@ -1,6 +1,6 @@
 #if !defined(ANDROID_r2_2_0) && !defined(ANDROID_r2_3_3) && !defined(ANDROID_r3_0_1) && \
  !defined(ANDROID_r4_0_0) && !defined(ANDROID_r4_0_3) && !defined(ANDROID_r4_1_1) && \
- !defined(ANDROID_r4_2_0) && !defined(ANDROID_r4_3_0)
+ !defined(ANDROID_r4_2_0) && !defined(ANDROID_r4_3_0) && !defined(ANDROID_r4_4_0)
 # error Building camera wrapper for your version of Android is not supported by OpenCV.\
  You need to modify OpenCV sources in order to compile camera wrapper for your version of Android.
 #endif
@@ -22,7 +22,7 @@
 #elif defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0)
 # include <gui/ISurface.h>
 # include <gui/BufferQueue.h>
-#elif defined(ANDROID_r4_3_0)
+#elif defined(ANDROID_r4_3_0) || defined(ANDROID_r4_4_0)
 # include <gui/IGraphicBufferProducer.h>
 # include <gui/BufferQueue.h>
 #else
@@ -74,6 +74,20 @@ public:
     {
     }
 };
+#elif defined(ANDROID_r4_4_0)
+class ConsumerListenerStub: public android::BnConsumerListener
+{
+public:
+    virtual void onFrameAvailable()
+    {
+    }
+    virtual void onBuffersReleased()
+    {
+    }
+    virtual ~ConsumerListenerStub()
+    {
+    }
+};
 #endif
 
 std::string getProcessName()
@@ -306,7 +320,8 @@ public:
     }
 
     virtual void postData(int32_t msgType, const sp<IMemory>& dataPtr
-    #if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) || defined(ANDROID_r4_3_0)
+#if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) \
+ || defined(ANDROID_r4_3_0) || defined(ANDROID_r4_4_0)
                           ,camera_frame_metadata_t*
 #endif
                           )
@@ -623,7 +638,14 @@ CameraHandler* CameraHandler::initCameraConnect(const CameraCallback& callback,
     bufferStatus = camera->setPreviewTexture(bufferQueue);
     if (bufferStatus != 0)
         LOGE("initCameraConnect: failed setPreviewTexture call; camera might not work correctly");
-#endif
+# elif defined(ANDROID_r4_4_0)
+    sp<BufferQueue> bufferQueue = new BufferQueue();
+    sp<IConsumerListener> queueListener = new ConsumerListenerStub();
+    bufferQueue->consumerConnect(queueListener, true);
+    bufferStatus = handler->camera->setPreviewTarget(bufferQueue);
+    if (bufferStatus != 0)
+        LOGE("applyProperties: failed setPreviewTexture call; camera might not work correctly");
+# endif
 
 #if (defined(ANDROID_r2_2_0) || defined(ANDROID_r2_3_3) || defined(ANDROID_r3_0_1))
 # if 1
@@ -663,7 +685,8 @@ void CameraHandler::closeCameraConnect()
     }
 
     camera->stopPreview();
-#if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) || defined(ANDROID_r4_3_0)
+#if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) \
+ || defined(ANDROID_r4_3_0) || defined(ANDROID_r4_3_0)
     camera->setPreviewCallbackFlags(CAMERA_FRAME_CALLBACK_FLAG_NOOP);
 #endif
     camera->disconnect();
@@ -914,7 +937,8 @@ void CameraHandler::applyProperties(CameraHandler** ppcameraHandler)
 
     CameraParameters curCameraParameters((*ppcameraHandler)->params.flatten());
 
-#if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) || defined(ANDROID_r4_3_0)
+#if defined(ANDROID_r4_0_0) || defined(ANDROID_r4_0_3) || defined(ANDROID_r4_1_1) || defined(ANDROID_r4_2_0) \
+ || defined(ANDROID_r4_3_0) || defined(ANDROID_r4_4_0)
     CameraHandler* handler=*ppcameraHandler;
 
     handler->camera->stopPreview();
@@ -943,6 +967,13 @@ void CameraHandler::applyProperties(CameraHandler** ppcameraHandler)
     bufferStatus = handler->camera->setPreviewTexture(bufferQueue);
     if (bufferStatus != 0)
         LOGE("applyProperties: failed setPreviewTexture call; camera might not work correctly");
+# elif defined(ANDROID_r4_4_0)
+    sp<BufferQueue> bufferQueue = new BufferQueue();
+    sp<IConsumerListener> queueListener = new ConsumerListenerStub();
+    bufferQueue->consumerConnect(queueListener, true);
+    bufferStatus = handler->camera->setPreviewTarget(bufferQueue);
+    if (bufferStatus != 0)
+        LOGE("applyProperties: failed setPreviewTexture call; camera might not work correctly");
 # endif
 
     handler->camera->setPreviewCallbackFlags( CAMERA_FRAME_CALLBACK_FLAG_ENABLE_MASK | CAMERA_FRAME_CALLBACK_FLAG_COPY_OUT_MASK);//with copy

diff --git a/modules/calib3d/src/stereosgbm.cpp b/modules/calib3d/src/stereosgbm.cpp
index 8caab41..0de9189 100644 (file)
--- a/modules/calib3d/src/stereosgbm.cpp
+++ b/modules/calib3d/src/stereosgbm.cpp
@@ -114,7 +114,7 @@ static void calcPixelCostBT( const Mat& img1, const Mat& img2, int y,
                             int tabOfs, int )
 {
     int x, c, width = img1.cols, cn = img1.channels();
-    int minX1 = max(maxD, 0), maxX1 = width + min(minD, 0);
+    int minX1 = max(-maxD, 0), maxX1 = width + min(minD, 0);
     int minX2 = max(minX1 - maxD, 0), maxX2 = min(maxX1 - minD, width);
     int D = maxD - minD, width1 = maxX1 - minX1, width2 = maxX2 - minX2;
     const PixType *row1 = img1.ptr<PixType>(y), *row2 = img2.ptr<PixType>(y);
@@ -329,7 +329,7 @@ static void computeDisparitySGBM( const Mat& img1, const Mat& img2,
     int disp12MaxDiff = params.disp12MaxDiff > 0 ? params.disp12MaxDiff : 1;
     int P1 = params.P1 > 0 ? params.P1 : 2, P2 = max(params.P2 > 0 ? params.P2 : 5, P1+1);
     int k, width = disp1.cols, height = disp1.rows;
-    int minX1 = max(maxD, 0), maxX1 = width + min(minD, 0);
+    int minX1 = max(-maxD, 0), maxX1 = width + min(minD, 0);
     int D = maxD - minD, width1 = maxX1 - minX1;
     int INVALID_DISP = minD - 1, INVALID_DISP_SCALED = INVALID_DISP*DISP_SCALE;
     int SW2 = SADWindowSize.width/2, SH2 = SADWindowSize.height/2;

diff --git a/modules/contrib/src/detection_based_tracker.cpp b/modules/contrib/src/detection_based_tracker.cpp
index 0b3fcce..381bde0 100644 (file)
--- a/modules/contrib/src/detection_based_tracker.cpp
+++ b/modules/contrib/src/detection_based_tracker.cpp
@@ -169,7 +169,6 @@ bool DetectionBasedTracker::SeparateDetectionWork::run()
     return true;
 }
 
-#ifdef __GNUC__
 #define CATCH_ALL_AND_LOG(_block)                                                           \
 do {                                                                                        \
     try {                                                                                   \
@@ -177,29 +176,13 @@ do {
         break;                                                                              \
     }                                                                                       \
     catch(cv::Exception& e) {                                                               \
-        LOGE0("\n %s: ERROR: OpenCV Exception caught: \n'%s'\n\n", __func__, e.what());     \
+        LOGE0("\n %s: ERROR: OpenCV Exception caught: \n'%s'\n\n", CV_Func, e.what());     \
     } catch(std::exception& e) {                                                            \
-        LOGE0("\n %s: ERROR: Exception caught: \n'%s'\n\n", __func__, e.what());            \
+        LOGE0("\n %s: ERROR: Exception caught: \n'%s'\n\n", CV_Func, e.what());            \
     } catch(...) {                                                                          \
-        LOGE0("\n %s: ERROR: UNKNOWN Exception caught\n\n", __func__);                      \
+        LOGE0("\n %s: ERROR: UNKNOWN Exception caught\n\n", CV_Func);                      \
     }                                                                                       \
 } while(0)
-#else
-#define CATCH_ALL_AND_LOG(_block)                                                           \
-do {                                                                                        \
-    try {                                                                                   \
-        _block;                                                                             \
-        break;                                                                              \
-    }                                                                                       \
-    catch(cv::Exception& e) {                                                               \
-        LOGE0("\n ERROR: OpenCV Exception caught: \n'%s'\n\n", e.what());                   \
-    } catch(std::exception& e) {                                                            \
-        LOGE0("\n ERROR: Exception caught: \n'%s'\n\n", e.what());                          \
-    } catch(...) {                                                                          \
-        LOGE0("\n ERROR: UNKNOWN Exception caught\n\n");                                    \
-    }                                                                                       \
-} while(0)
-#endif
 
 void* workcycleObjectDetectorFunction(void* p)
 {

diff --git a/modules/core/include/opencv2/core/core.hpp b/modules/core/include/opencv2/core/core.hpp
index af3a50c..b6426d8 100644 (file)
--- a/modules/core/include/opencv2/core/core.hpp
+++ b/modules/core/include/opencv2/core/core.hpp
@@ -164,7 +164,7 @@ public:
 
     int code; ///< error code @see CVStatus
     string err; ///< error description
-    string func; ///< function name. Available only when the compiler supports __func__ macro
+    string func; ///< function name. Available only when the compiler supports getting it
     string file; ///< source file name where the error has occured
     int line; ///< line number in the source file where the error has occured
 };
@@ -209,16 +209,19 @@ typedef int (CV_CDECL *ErrorCallback)( int status, const char* func_name,
 CV_EXPORTS ErrorCallback redirectError( ErrorCallback errCallback,
                                         void* userdata=0, void** prevUserdata=0);
 
-#ifdef __GNUC__
-#define CV_Error( code, msg ) cv::error( cv::Exception(code, msg, __func__, __FILE__, __LINE__) )
-#define CV_Error_( code, args ) cv::error( cv::Exception(code, cv::format args, __func__, __FILE__, __LINE__) )
-#define CV_Assert( expr ) if(!!(expr)) ; else cv::error( cv::Exception(CV_StsAssert, #expr, __func__, __FILE__, __LINE__) )
+
+#if defined __GNUC__
+#define CV_Func __func__
+#elif defined _MSC_VER
+#define CV_Func __FUNCTION__
 #else
-#define CV_Error( code, msg ) cv::error( cv::Exception(code, msg, "", __FILE__, __LINE__) )
-#define CV_Error_( code, args ) cv::error( cv::Exception(code, cv::format args, "", __FILE__, __LINE__) )
-#define CV_Assert( expr ) if(!!(expr)) ; else cv::error( cv::Exception(CV_StsAssert, #expr, "", __FILE__, __LINE__) )
+#define CV_Func ""
 #endif
 
+#define CV_Error( code, msg ) cv::error( cv::Exception(code, msg, CV_Func, __FILE__, __LINE__) )
+#define CV_Error_( code, args ) cv::error( cv::Exception(code, cv::format args, CV_Func, __FILE__, __LINE__) )
+#define CV_Assert( expr ) if(!!(expr)) ; else cv::error( cv::Exception(CV_StsAssert, #expr, CV_Func, __FILE__, __LINE__) )
+
 #ifdef _DEBUG
 #define CV_DbgAssert(expr) CV_Assert(expr)
 #else

diff --git a/modules/core/include/opencv2/core/internal.hpp b/modules/core/include/opencv2/core/internal.hpp
index 2f26e7c..3cd2f90 100644 (file)
--- a/modules/core/include/opencv2/core/internal.hpp
+++ b/modules/core/include/opencv2/core/internal.hpp
@@ -774,11 +774,7 @@ namespace cv { namespace ogl {
 CV_EXPORTS bool checkError(const char* file, const int line, const char* func = "");
 }}
 
-#if defined(__GNUC__)
-    #define CV_CheckGlError() CV_DbgAssert( (cv::ogl::checkError(__FILE__, __LINE__, __func__)) )
-#else
-    #define CV_CheckGlError() CV_DbgAssert( (cv::ogl::checkError(__FILE__, __LINE__)) )
-#endif
+#define CV_CheckGlError() CV_DbgAssert( (cv::ogl::checkError(__FILE__, __LINE__, CV_Func)) )
 
 #endif //__cplusplus
 

diff --git a/modules/core/include/opencv2/core/mat.hpp b/modules/core/include/opencv2/core/mat.hpp
index 4cb12cc..8ddc16e 100644 (file)
--- a/modules/core/include/opencv2/core/mat.hpp
+++ b/modules/core/include/opencv2/core/mat.hpp
@@ -1075,7 +1075,9 @@ template<typename _Tp> template<int n> inline Mat_<_Tp>::operator Vec<typename D
 template<typename _Tp> template<int m, int n> inline Mat_<_Tp>::operator Matx<typename DataType<_Tp>::channel_type, m, n>() const
 {
     CV_Assert(n % DataType<_Tp>::channels == 0);
-    return this->Mat::operator Matx<typename DataType<_Tp>::channel_type, m, n>();
+
+    Matx<typename DataType<_Tp>::channel_type, m, n> res = this->Mat::operator Matx<typename DataType<_Tp>::channel_type, m, n>();
+    return res;
 }
 
 template<typename T1, typename T2, typename Op> inline void

diff --git a/modules/core/src/gpumat.cpp b/modules/core/src/gpumat.cpp
index 53e118a..ff459f9 100644 (file)
--- a/modules/core/src/gpumat.cpp
+++ b/modules/core/src/gpumat.cpp
@@ -72,13 +72,8 @@ using namespace cv::gpu;
 
 namespace
 {
-#if defined(__GNUC__)
-    #define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__, __func__)
-    #define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__, __func__)
-#else /* defined(__CUDACC__) || defined(__MSVC__) */
-    #define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__)
-    #define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__)
-#endif
+#define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__, CV_Func)
+#define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__, CV_Func)
 
     inline void ___cudaSafeCall(cudaError_t err, const char *file, const int line, const char *func = "")
     {

diff --git a/modules/core/src/opengl_interop.cpp b/modules/core/src/opengl_interop.cpp
index 0cb9d3d..bf368ea 100644 (file)
--- a/modules/core/src/opengl_interop.cpp
+++ b/modules/core/src/opengl_interop.cpp
@@ -69,11 +69,7 @@ namespace
         #else
             void throw_nocuda() { CV_Error(CV_StsNotImplemented, "The called functionality is disabled for current build or platform"); }
 
-            #if defined(__GNUC__)
-                #define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__, __func__)
-            #else /* defined(__CUDACC__) || defined(__MSVC__) */
-                #define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__)
-            #endif
+            #define cudaSafeCall(expr)  ___cudaSafeCall(expr, __FILE__, __LINE__, CV_Func)
 
             void ___cudaSafeCall(cudaError_t err, const char* file, const int line, const char* func = "")
             {
@@ -1026,6 +1022,7 @@ cv::ogl::Texture2D::Texture2D(InputArray arr, bool autoRelease) : rows_(0), cols
             #else
                 GpuMat dmat = arr.getGpuMat();
                 ogl::Buffer buf(dmat, ogl::Buffer::PIXEL_UNPACK_BUFFER);
+                buf.setAutoRelease(true);
                 buf.bind(ogl::Buffer::PIXEL_UNPACK_BUFFER);
                 impl_ = new Impl(internalFormats[cn], asize.width, asize.height, srcFormats[cn], gl_types[depth], 0, autoRelease);
                 ogl::Buffer::unbind(ogl::Buffer::PIXEL_UNPACK_BUFFER);
@@ -1139,6 +1136,7 @@ void cv::ogl::Texture2D::copyFrom(InputArray arr, bool autoRelease)
             #else
                 GpuMat dmat = arr.getGpuMat();
                 ogl::Buffer buf(dmat, ogl::Buffer::PIXEL_UNPACK_BUFFER);
+                buf.setAutoRelease(true);
                 buf.bind(ogl::Buffer::PIXEL_UNPACK_BUFFER);
                 impl_->copyFrom(asize.width, asize.height, srcFormats[cn], gl_types[depth], 0);
                 ogl::Buffer::unbind(ogl::Buffer::PIXEL_UNPACK_BUFFER);
@@ -1189,6 +1187,7 @@ void cv::ogl::Texture2D::copyTo(OutputArray arr, int ddepth, bool autoRelease) c
                 throw_nocuda();
             #else
                 ogl::Buffer buf(rows_, cols_, CV_MAKE_TYPE(ddepth, cn), ogl::Buffer::PIXEL_PACK_BUFFER);
+                buf.setAutoRelease(true);
                 buf.bind(ogl::Buffer::PIXEL_PACK_BUFFER);
                 impl_->copyTo(dstFormat, gl_types[ddepth], 0);
                 ogl::Buffer::unbind(ogl::Buffer::PIXEL_PACK_BUFFER);

diff --git a/modules/core/src/persistence.cpp b/modules/core/src/persistence.cpp
index bf6a64c..7759f70 100644 (file)
--- a/modules/core/src/persistence.cpp
+++ b/modules/core/src/persistence.cpp
@@ -415,13 +415,8 @@ cvCreateMap( int flags, int header_size, int elem_size,
     return map;
 }
 
-#ifdef __GNUC__
 #define CV_PARSE_ERROR( errmsg )                                    \
-    icvParseError( fs, __func__, (errmsg), __FILE__, __LINE__ )
-#else
-#define CV_PARSE_ERROR( errmsg )                                    \
-    icvParseError( fs, "", (errmsg), __FILE__, __LINE__ )
-#endif
+    icvParseError( fs, CV_Func, (errmsg), __FILE__, __LINE__ )
 
 static void
 icvParseError( CvFileStorage* fs, const char* func_name,

diff --git a/modules/gpu/src/cuda/safe_call.hpp b/modules/gpu/src/cuda/safe_call.hpp
index 53d691f..2eeaf85 100644 (file)
--- a/modules/gpu/src/cuda/safe_call.hpp
+++ b/modules/gpu/src/cuda/safe_call.hpp
@@ -81,11 +81,7 @@ static inline void ___nppSafeCall(int err, const char *file, const int line, con
         cv::gpu::nppError(err, file, line, func);
 }
 
-#if defined(__GNUC__)
-    #define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__, __func__)
-#else
-    #define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__)
-#endif
+#define nppSafeCall(expr)  ___nppSafeCall(expr, __FILE__, __LINE__, CV_Func)
 
 // ncvSafeCall
 
@@ -95,11 +91,7 @@ static inline void ___ncvSafeCall(int err, const char *file, const int line, con
         cv::gpu::ncvError(err, file, line, func);
 }
 
-#if defined(__GNUC__)
-    #define ncvSafeCall(expr)  ___ncvSafeCall(expr, __FILE__, __LINE__, __func__)
-#else
-    #define ncvSafeCall(expr)  ___ncvSafeCall(expr, __FILE__, __LINE__)
-#endif
+#define ncvSafeCall(expr)  ___ncvSafeCall(expr, __FILE__, __LINE__, CV_Func)
 
 // cufftSafeCall
 
@@ -110,11 +102,7 @@ static inline void ___ncvSafeCall(int err, const char *file, const int line, con
             cv::gpu::cufftError(err, file, line, func);
     }
 
-    #if defined(__GNUC__)
-        #define cufftSafeCall(expr)  ___cufftSafeCall(expr, __FILE__, __LINE__, __func__)
-    #else
-        #define cufftSafeCall(expr)  ___cufftSafeCall(expr, __FILE__, __LINE__)
-    #endif
+#define cufftSafeCall(expr)  ___cufftSafeCall(expr, __FILE__, __LINE__, CV_Func)
 #endif
 
 // cublasSafeCall
@@ -126,11 +114,7 @@ static inline void ___ncvSafeCall(int err, const char *file, const int line, con
             cv::gpu::cublasError(err, file, line, func);
     }
 
-    #if defined(__GNUC__)
-        #define cublasSafeCall(expr)  ___cublasSafeCall(expr, __FILE__, __LINE__, __func__)
-    #else
-        #define cublasSafeCall(expr)  ___cublasSafeCall(expr, __FILE__, __LINE__)
-    #endif
+#define cublasSafeCall(expr)  ___cublasSafeCall(expr, __FILE__, __LINE__, CV_Func)
 #endif
 
 #endif /* __OPENCV_CUDA_SAFE_CALL_HPP__ */

diff --git a/modules/highgui/src/cap_ffmpeg_impl.hpp b/modules/highgui/src/cap_ffmpeg_impl.hpp
index 2d8ad22..151a0ca 100644 (file)
--- a/modules/highgui/src/cap_ffmpeg_impl.hpp
+++ b/modules/highgui/src/cap_ffmpeg_impl.hpp
@@ -546,7 +546,7 @@ bool CvCapture_FFMPEG::open( const char* _filename )
         goto exit_func;
     }
     err =
-#if LIBAVFORMAT_BUILD >= CALC_FFMPEG_VERSION(53, 3, 0)
+#if LIBAVFORMAT_BUILD >= CALC_FFMPEG_VERSION(53, 6, 0)
     avformat_find_stream_info(ic, NULL);
 #else
     av_find_stream_info(ic);
@@ -2114,7 +2114,7 @@ bool InputMediaStream_FFMPEG::open(const char* fileName, int* codec, int* chroma
     if (err < 0)
         return false;
 
-    #if LIBAVFORMAT_BUILD >= CALC_FFMPEG_VERSION(53, 3, 0)
+    #if LIBAVFORMAT_BUILD >= CALC_FFMPEG_VERSION(53, 6, 0)
         err = avformat_find_stream_info(ctx_, 0);
     #else
         err = av_find_stream_info(ctx_);

diff --git a/modules/highgui/src/cap_images.cpp b/modules/highgui/src/cap_images.cpp
index 4cf51d8..e1a8b8b 100644 (file)
--- a/modules/highgui/src/cap_images.cpp
+++ b/modules/highgui/src/cap_images.cpp
@@ -200,10 +200,20 @@ static char* icvExtractPattern(const char *filename, unsigned *offset)
     }
     else // no pattern filename was given - extract the pattern
     {
-        for(at = name; *at && !isdigit(*at); at++)
-            ;
+        at = name;
+
+        // ignore directory names
+        char *slash = strrchr(at, '/');
+        if (slash) at = slash + 1;
+
+#ifdef _WIN32
+        slash = strrchr(at, '\\');
+        if (slash) at = slash + 1;
+#endif
+
+        while (*at && !isdigit(*at)) at++;
 
-        if(!at)
+        if(!*at)
             return 0;
 
         sscanf(at, "%u", offset);

diff --git a/modules/highgui/src/cap_openni.cpp b/modules/highgui/src/cap_openni.cpp
index 5187712..402a0cd 100644 (file)
--- a/modules/highgui/src/cap_openni.cpp
+++ b/modules/highgui/src/cap_openni.cpp
@@ -172,6 +172,9 @@ public:
     xn::ImageGenerator &imageGenerator;
 
 private:
+    ApproximateSyncGrabber(const ApproximateSyncGrabber&);
+    ApproximateSyncGrabber& operator=(ApproximateSyncGrabber&);
+
     int maxBufferSize;
     bool isCircleBuffer;
     int maxTimeDuration;
@@ -215,7 +218,7 @@ private:
         virtual bool grab( xn::DepthMetaData& depthMetaData,
                            xn::ImageMetaData& imageMetaData )
         {
-            while(1)
+            for(;;)
             {
                 if( !isDepthFilled )
                     isDepthFilled = popDepthMetaData(depth);
@@ -971,7 +974,7 @@ double CvCapture_OpenNI::getDepthGeneratorProperty( int propIdx )
         propValue = depthGenerator.GetAlternativeViewPointCap().IsViewPointAs(imageGenerator) ? 1.0 : 0.0;
         break;
     case CV_CAP_PROP_POS_MSEC :
-        propValue = depthGenerator.GetTimestamp();
+        propValue = (double)depthGenerator.GetTimestamp();
         break;
     case CV_CAP_PROP_POS_FRAMES :
         propValue = depthGenerator.GetFrameID();
@@ -1067,10 +1070,10 @@ double CvCapture_OpenNI::getImageGeneratorProperty( int propIdx )
             propValue = mode.nFPS;
         break;
     case CV_CAP_PROP_POS_MSEC :
-        propValue = imageGenerator.GetTimestamp();
+        propValue = (double)imageGenerator.GetTimestamp();
         break;
     case CV_CAP_PROP_POS_FRAMES :
-        propValue = imageGenerator.GetFrameID();
+        propValue = (double)imageGenerator.GetFrameID();
         break;
     default :
     {

diff --git a/modules/highgui/test/test_ffmpeg.cpp b/modules/highgui/test/test_ffmpeg.cpp
index 468fe77..30410ea 100644 (file)
--- a/modules/highgui/test/test_ffmpeg.cpp
+++ b/modules/highgui/test/test_ffmpeg.cpp
@@ -154,7 +154,7 @@ public:
     {
         try
         {
-            string filename = ts->get_data_path() + "../cv/features2d/tsukuba.png";
+            string filename = ts->get_data_path() + "readwrite/ordinary.bmp";
             VideoCapture cap(filename);
             Mat img0 = imread(filename, 1);
             Mat img, img_next;

diff --git a/modules/java/CMakeLists.txt b/modules/java/CMakeLists.txt
index 10bb559..5012f91 100644 (file)
--- a/modules/java/CMakeLists.txt
+++ b/modules/java/CMakeLists.txt
@@ -150,9 +150,10 @@ endforeach()
 set(step2_depends ${step1_depends} ${scripts_gen_javadoc} ${scripts_rst_parser} ${javadoc_rst_sources} ${generated_java_sources} ${handwrittren_java_sources})
 string(REPLACE ";" "," OPENCV_JAVA_MODULES_STR "${OPENCV_JAVA_MODULES}")
 add_custom_command(OUTPUT ${documented_java_files}
-                   COMMAND ${PYTHON_EXECUTABLE} "${scripts_gen_javadoc}" --modules ${OPENCV_JAVA_MODULES_STR} "${CMAKE_CURRENT_SOURCE_DIR}/generator/src/java" "${CMAKE_CURRENT_BINARY_DIR}" 2>"${CMAKE_CURRENT_BINARY_DIR}/get_javadoc_errors.log"
+                   COMMAND ${PYTHON_EXECUTABLE} "${scripts_gen_javadoc}" --modules ${OPENCV_JAVA_MODULES_STR} "${CMAKE_CURRENT_SOURCE_DIR}/generator/src/java" "${CMAKE_CURRENT_BINARY_DIR}" 2> "${CMAKE_CURRENT_BINARY_DIR}/get_javadoc_errors.log"
                    WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
                    DEPENDS ${step2_depends}
+                   VERBATIM
                   )
 
 # step 3: copy files to destination

diff --git a/modules/ocl/doc/image_processing.rst b/modules/ocl/doc/image_processing.rst
index 7dde475..100876a 100644 (file)
--- a/modules/ocl/doc/image_processing.rst
+++ b/modules/ocl/doc/image_processing.rst
@@ -65,15 +65,15 @@ ocl::integral
 -----------------
 Computes an integral image.
 
-.. ocv:function:: void ocl::integral(const oclMat &src, oclMat &sum, oclMat &sqsum)
+.. ocv:function:: void ocl::integral(const oclMat &src, oclMat &sum, oclMat &sqsum, int sdepth=-1)
 
-.. ocv:function:: void ocl::integral(const oclMat &src, oclMat &sum)
+.. ocv:function:: void ocl::integral(const oclMat &src, oclMat &sum, int sdepth=-1)
 
     :param src: Source image. Only  ``CV_8UC1`` images are supported for now.
 
-    :param sum: Integral image containing 32-bit unsigned integer values packed into  ``CV_32SC1`` .
+    :param sum: Integral image containing 32-bit unsigned integer or 32-bit floating-point .
 
-    :param sqsum: Sqsum values is ``CV_32FC1`` type.
+    :param sqsum: Sqsum values is ``CV_32FC1`` or ``CV_64FC1`` type.
 
 .. seealso:: :ocv:func:`integral`
 

diff --git a/modules/ocl/doc/ml_machine_learning.rst b/modules/ocl/doc/ml_machine_learning.rst
index eb72cbe..ad0e303 100644 (file)
--- a/modules/ocl/doc/ml_machine_learning.rst
+++ b/modules/ocl/doc/ml_machine_learning.rst
@@ -91,11 +91,7 @@ ocl::distanceToCenters
 ----------------------
 For each samples in ``source``, find its closest neighour in ``centers``.
 
-.. ocv:function:: void ocl::distanceToCenters(oclMat &dists, oclMat &labels, const oclMat &src, const oclMat &centers, int distType = NORM_L2SQR, const oclMat &indices = oclMat())
-
-    :param dists: The output distances calculated from each sample to the best matched center.
-
-    :param labels: The output index of best matched center for each row of sample.
+.. ocv:function:: void ocl::distanceToCenters(const oclMat &src, const oclMat &centers, Mat &dists, Mat &labels, int distType = NORM_L2SQR)
 
     :param src: Floating-point matrix of input samples. One row per sample.
 
@@ -103,10 +99,8 @@ For each samples in ``source``, find its closest neighour in ``centers``.
 
     :param distType: Distance metric to calculate distances. Supports ``NORM_L1`` and ``NORM_L2SQR``.
 
-    :param indices: Optional source indices. If not empty:
+    :param dists: The output distances calculated from each sample to the best matched center.
 
-            * only the indexed source samples will be processed
-            * outputs, i.e., ``dists`` and ``labels``, have the same size of indices
-            * outputs are in the same order of indices instead of the order of src
+    :param labels: The output index of best matched center for each row of sample.
 
 The method is a utility function which maybe used for multiple clustering algorithms such as K-means.

diff --git a/modules/ocl/include/opencv2/ocl/ocl.hpp b/modules/ocl/include/opencv2/ocl/ocl.hpp
index fa88251..b8c9b85 100644 (file)
--- a/modules/ocl/include/opencv2/ocl/ocl.hpp
+++ b/modules/ocl/include/opencv2/ocl/ocl.hpp
@@ -861,10 +861,10 @@ namespace cv
         CV_EXPORTS void warpPerspective(const oclMat &src, oclMat &dst, const Mat &M, Size dsize, int flags = INTER_LINEAR);
 
         //! computes the integral image and integral for the squared image
-        // sum will have CV_32S type, sqsum - CV32F type
+        // sum will support CV_32S, CV_32F, sqsum - support CV32F, CV_64F
         // supports only CV_8UC1 source type
-        CV_EXPORTS void integral(const oclMat &src, oclMat &sum, oclMat &sqsum);
-        CV_EXPORTS void integral(const oclMat &src, oclMat &sum);
+        CV_EXPORTS void integral(const oclMat &src, oclMat &sum, oclMat &sqsum, int sdepth=-1 );
+        CV_EXPORTS void integral(const oclMat &src, oclMat &sum, int sdepth=-1 );
         CV_EXPORTS void cornerHarris(const oclMat &src, oclMat &dst, int blockSize, int ksize, double k, int bordertype = cv::BORDER_DEFAULT);
         CV_EXPORTS void cornerHarris_dxdy(const oclMat &src, oclMat &dst, oclMat &Dx, oclMat &Dy,
             int blockSize, int ksize, double k, int bordertype = cv::BORDER_DEFAULT);
@@ -879,7 +879,7 @@ namespace cv
         // supports NORM_L1 and NORM_L2 distType
         // if indices is provided, only the indexed rows will be calculated and their results are in the same
         // order of indices
-        CV_EXPORTS void distanceToCenters(oclMat &dists, oclMat &labels, const oclMat &src, const oclMat &centers, int distType = NORM_L2SQR, const oclMat &indices = oclMat());
+        CV_EXPORTS void distanceToCenters(const oclMat &src, const oclMat &centers, Mat &dists, Mat &labels, int distType = NORM_L2SQR);
 
         //!Does k-means procedure on GPU
         // supports CV_32FC1/CV_32FC2/CV_32FC4 data type
@@ -939,7 +939,7 @@ namespace cv
             Size m_maxSize;
             vector<CvSize> sizev;
             vector<float> scalev;
-            oclMat gimg1, gsum, gsqsum;
+            oclMat gimg1, gsum, gsqsum, gsqsum_t;
             void * buffers;
         };
 

diff --git a/modules/ocl/include/opencv2/ocl/private/opencl_dumpinfo.hpp b/modules/ocl/include/opencv2/ocl/private/opencl_dumpinfo.hpp
index beb3d27..a1b5300 100644 (file)
--- a/modules/ocl/include/opencv2/ocl/private/opencl_dumpinfo.hpp
+++ b/modules/ocl/include/opencv2/ocl/private/opencl_dumpinfo.hpp
@@ -39,16 +39,16 @@
 //
 //M*/
 
-#if !defined(DUMP_INFO_STDOUT) && !defined(DUMP_INFO_XML)
+#if !defined(DUMP_MESSAGE_STDOUT) && !defined(DUMP_PROPERTY_XML)
 #error Invalid usage
 #endif
 
-#if !defined(DUMP_INFO_STDOUT)
-#define DUMP_INFO_STDOUT(...)
+#if !defined(DUMP_PROPERTY_XML)
+#define DUMP_PROPERTY_XML(...)
 #endif
 
-#if !defined(DUMP_INFO_XML)
-#define DUMP_INFO_XML(...)
+#if !defined(DUMP_MESSAGE_STDOUT)
+#define DUMP_MESSAGE_STDOUT(...)
 #endif
 
 #include <sstream>
@@ -85,51 +85,70 @@ static void dumpOpenCLDevice()
     using namespace cv::ocl;
     try
     {
+        cv::ocl::PlatformsInfo platforms;
+        cv::ocl::getOpenCLPlatforms(platforms);
+        DUMP_MESSAGE_STDOUT("OpenCL Platforms: ");
+        const char* deviceTypeStr;
+        for(unsigned int i=0; i < platforms.size(); i++)
+        {
+            DUMP_MESSAGE_STDOUT("    " << platforms.at(i)->platformName);
+            const cv::ocl::DevicesInfo& devices = platforms.at(i)->devices;
+            for(unsigned int j=0; j < devices.size(); j++)
+            {
+                const cv::ocl::DeviceInfo& current_device = *devices.at(j);
+                deviceTypeStr = current_device.deviceType == CVCL_DEVICE_TYPE_CPU
+                            ? ("CPU") : (current_device.deviceType == CVCL_DEVICE_TYPE_GPU ? "GPU" : "unknown");
+                DUMP_MESSAGE_STDOUT( "        " << deviceTypeStr << " : " << current_device.deviceName << " : " << current_device.deviceVersion );
+                DUMP_PROPERTY_XML("cv_ocl_platform_"<< i<<"_device_"<<j, "(Platform=" << current_device.platform->platformName << ")(Type="
+                    << deviceTypeStr <<")(Name="<< current_device.deviceName << ")(Version="<< current_device.deviceVersion<<")");
+            }
+        }
+        DUMP_MESSAGE_STDOUT("Current OpenCL device: ");
+
         const cv::ocl::DeviceInfo& deviceInfo = cv::ocl::Context::getContext()->getDeviceInfo();
 
-        const char* deviceTypeStr = deviceInfo.deviceType == CVCL_DEVICE_TYPE_CPU
-                        ? "CPU" :
-                        (deviceInfo.deviceType == CVCL_DEVICE_TYPE_GPU ? "GPU" : "unknown");
-        DUMP_INFO_STDOUT("Device type", deviceTypeStr);
-        DUMP_INFO_XML("cv_ocl_deviceType", deviceTypeStr);
+        DUMP_MESSAGE_STDOUT("    Platform = "<< deviceInfo.platform->platformName);
+        DUMP_PROPERTY_XML("cv_ocl_current_platformName", deviceInfo.platform->platformName);
 
-        DUMP_INFO_STDOUT("Platform name", deviceInfo.platform->platformName);
-        DUMP_INFO_XML("cv_ocl_platformName", deviceInfo.platform->platformName);
+        deviceTypeStr = deviceInfo.deviceType == CVCL_DEVICE_TYPE_CPU
+                        ? "CPU" : (deviceInfo.deviceType == CVCL_DEVICE_TYPE_GPU ? "GPU" : "unknown");
+        DUMP_MESSAGE_STDOUT("    Type = "<< deviceTypeStr);
+        DUMP_PROPERTY_XML("cv_ocl_current_deviceType", deviceTypeStr);
 
-        DUMP_INFO_STDOUT("Device name", deviceInfo.deviceName);
-        DUMP_INFO_XML("cv_ocl_deviceName", deviceInfo.deviceName);
+        DUMP_MESSAGE_STDOUT("    Name = "<< deviceInfo.deviceName);
+        DUMP_PROPERTY_XML("cv_ocl_current_deviceName", deviceInfo.deviceName);
 
-        DUMP_INFO_STDOUT("Device version", deviceInfo.deviceVersion);
-        DUMP_INFO_XML("cv_ocl_deviceVersion", deviceInfo.deviceVersion);
+        DUMP_MESSAGE_STDOUT("    Version = " << deviceInfo.deviceVersion);
+        DUMP_PROPERTY_XML("cv_ocl_current_deviceVersion", deviceInfo.deviceVersion);
 
-        DUMP_INFO_STDOUT("Compute units", deviceInfo.maxComputeUnits);
-        DUMP_INFO_XML("cv_ocl_maxComputeUnits", deviceInfo.maxComputeUnits);
+        DUMP_MESSAGE_STDOUT("    Compute units = "<< deviceInfo.maxComputeUnits);
+        DUMP_PROPERTY_XML("cv_ocl_current_maxComputeUnits", deviceInfo.maxComputeUnits);
 
-        DUMP_INFO_STDOUT("Max work group size", deviceInfo.maxWorkGroupSize);
-        DUMP_INFO_XML("cv_ocl_maxWorkGroupSize", deviceInfo.maxWorkGroupSize);
+        DUMP_MESSAGE_STDOUT("    Max work group size = "<< deviceInfo.maxWorkGroupSize);
+        DUMP_PROPERTY_XML("cv_ocl_current_maxWorkGroupSize", deviceInfo.maxWorkGroupSize);
 
         std::string localMemorySizeStr = bytesToStringRepr(deviceInfo.localMemorySize);
-        DUMP_INFO_STDOUT("Local memory size", localMemorySizeStr.c_str());
-        DUMP_INFO_XML("cv_ocl_localMemorySize", deviceInfo.localMemorySize);
+        DUMP_MESSAGE_STDOUT("    Local memory size = "<< localMemorySizeStr.c_str());
+        DUMP_PROPERTY_XML("cv_ocl_current_localMemorySize", deviceInfo.localMemorySize);
 
         std::string maxMemAllocSizeStr = bytesToStringRepr(deviceInfo.maxMemAllocSize);
-        DUMP_INFO_STDOUT("Max memory allocation size", maxMemAllocSizeStr.c_str());
-        DUMP_INFO_XML("cv_ocl_maxMemAllocSize", deviceInfo.maxMemAllocSize);
+        DUMP_MESSAGE_STDOUT("    Max memory allocation size = "<< maxMemAllocSizeStr.c_str());
+        DUMP_PROPERTY_XML("cv_ocl_current_maxMemAllocSize", deviceInfo.maxMemAllocSize);
 
         const char* doubleSupportStr = deviceInfo.haveDoubleSupport ? "Yes" : "No";
-        DUMP_INFO_STDOUT("Double support", doubleSupportStr);
-        DUMP_INFO_XML("cv_ocl_haveDoubleSupport", deviceInfo.haveDoubleSupport);
+        DUMP_MESSAGE_STDOUT("    Double support = "<< doubleSupportStr);
+        DUMP_PROPERTY_XML("cv_ocl_current_haveDoubleSupport", deviceInfo.haveDoubleSupport);
 
         const char* isUnifiedMemoryStr = deviceInfo.isUnifiedMemory ? "Yes" : "No";
-        DUMP_INFO_STDOUT("Unified memory", isUnifiedMemoryStr);
-        DUMP_INFO_XML("cv_ocl_isUnifiedMemory", deviceInfo.isUnifiedMemory);
+        DUMP_MESSAGE_STDOUT("    Unified memory = "<< isUnifiedMemoryStr);
+        DUMP_PROPERTY_XML("cv_ocl_current_isUnifiedMemory", deviceInfo.isUnifiedMemory);
     }
     catch (...)
     {
-        DUMP_INFO_STDOUT("OpenCL device", "not available");
-        DUMP_INFO_XML("cv_ocl", "not available");
+        DUMP_MESSAGE_STDOUT("OpenCL device not available");
+        DUMP_PROPERTY_XML("cv_ocl", "not available");
     }
 }
 
-#undef DUMP_INFO_STDOUT
-#undef DUMP_INFO_XML
+#undef DUMP_MESSAGE_STDOUT
+#undef DUMP_PROPERTY_XML

diff --git a/modules/ocl/perf/main.cpp b/modules/ocl/perf/main.cpp
index 836f8ee..5a67d1c 100644 (file)
--- a/modules/ocl/perf/main.cpp
+++ b/modules/ocl/perf/main.cpp
@@ -42,17 +42,20 @@
 
 #include "perf_precomp.hpp"
 
-#define DUMP_INFO_STDOUT(propertyDisplayName, propertyValue) \
+#define DUMP_PROPERTY_XML(propertyName, propertyValue) \
     do { \
-        std::cout << (propertyDisplayName) << ": " << (propertyValue) << std::endl; \
+        std::stringstream ssName, ssValue;\
+        ssName << propertyName;\
+        ssValue << propertyValue; \
+        ::testing::Test::RecordProperty(ssName.str(), ssValue.str()); \
     } while (false)
 
-#define DUMP_INFO_XML(propertyXMLName, propertyValue) \
+#define DUMP_MESSAGE_STDOUT(msg) \
     do { \
-        std::stringstream ss; ss << propertyValue; \
-        ::testing::Test::RecordProperty((propertyXMLName), ss.str()); \
+        std::cout << msg << std::endl; \
     } while (false)
 
+
 #include "opencv2/ocl/private/opencl_dumpinfo.hpp"
 
 static const char * impls[] =

diff --git a/modules/ocl/perf/perf_arithm.cpp b/modules/ocl/perf/perf_arithm.cpp
index 24eab3b..2699b44 100644 (file)
--- a/modules/ocl/perf/perf_arithm.cpp
+++ b/modules/ocl/perf/perf_arithm.cpp
@@ -1066,6 +1066,9 @@ PERF_TEST_P(RepeatFixture, Repeat,
     const int nx = 3, ny = 2;
     const Size dstSize(srcSize.width * nx, srcSize.height * ny);
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+    checkDeviceMaxMemoryAllocSize(dstSize, type);
+
     Mat src(srcSize, type), dst(dstSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 

diff --git a/modules/ocl/perf/perf_filters.cpp b/modules/ocl/perf/perf_filters.cpp
index b6dcd2a..7e5389d 100644 (file)
--- a/modules/ocl/perf/perf_filters.cpp
+++ b/modules/ocl/perf/perf_filters.cpp
@@ -61,6 +61,8 @@ PERF_TEST_P(BlurFixture, Blur,
     const Size srcSize = get<0>(params), ksize(3, 3);
     const int type = get<1>(params), bordertype = BORDER_CONSTANT;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -99,6 +101,8 @@ PERF_TEST_P(LaplacianFixture, Laplacian,
     const Size srcSize = get<0>(params);
     const int type = get<1>(params), ksize = 3;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -138,6 +142,8 @@ PERF_TEST_P(ErodeFixture, Erode,
     const int type = get<1>(params), ksize = 3;
     const Mat ker = getStructuringElement(MORPH_RECT, Size(ksize, ksize));
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst).in(ker);
 
@@ -176,6 +182,8 @@ PERF_TEST_P(SobelFixture, Sobel,
     const Size srcSize = get<0>(params);
     const int type = get<1>(params), dx = 1, dy = 1;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type, sizeof(float) * 2);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -217,6 +225,8 @@ PERF_TEST_P(ScharrFixture, Scharr,
     const Size srcSize = get<0>(params);
     const int type = get<1>(params), dx = 1, dy = 0;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type, sizeof(float) * 2);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -258,6 +268,8 @@ PERF_TEST_P(GaussianBlurFixture, GaussianBlur,
     const Size srcSize = get<0>(params);
     const int type = get<1>(params), ksize = 7;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -295,6 +307,8 @@ PERF_TEST_P(filter2DFixture, filter2D,
     const Size srcSize = get<0>(params);
     const int type = get<1>(params), ksize = 3;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type), kernel(ksize, ksize, CV_32SC1);
     declare.in(src, WARMUP_RNG).in(kernel).out(dst);
     randu(kernel, -3.0, 3.0);
@@ -335,6 +349,8 @@ PERF_TEST_P(BilateralFixture, Bilateral,
     const int type = get<1>(params), d = 7;
     const double sigmacolor = 50.0, sigmaspace = 50.0;
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -366,8 +382,7 @@ PERF_TEST_P(BilateralFixture, Bilateral,
 typedef Size_MatType adaptiveBilateralFixture;
 
 PERF_TEST_P(adaptiveBilateralFixture, adaptiveBilateral,
-            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
-                               OCL_PERF_ENUM(CV_8UC1, CV_8UC3)))
+            ::testing::Combine(::testing::Values(OCL_SIZE_1000), OCL_PERF_ENUM(CV_8UC1, CV_8UC3)))
 {
     const Size_MatType_t params = GetParam();
     const Size srcSize = get<0>(params);
@@ -375,14 +390,11 @@ PERF_TEST_P(adaptiveBilateralFixture, adaptiveBilateral,
     const double sigmaspace = 10.0;
     Size ksize(9, 9);
 
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
     Mat src(srcSize, type), dst(srcSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
-    if (srcSize == OCL_SIZE_4000)
-        declare.time(type == CV_8UC3 ? 46 : 28);
-    else if (srcSize == OCL_SIZE_2000)
-        declare.time(type == CV_8UC3 ? 11 : 7);
-
     if (RUN_OCL_IMPL)
     {
         ocl::oclMat oclSrc(src), oclDst(srcSize, type);

diff --git a/modules/ocl/perf/perf_imgproc.cpp b/modules/ocl/perf/perf_imgproc.cpp
index c57950f..e0ffe56 100644 (file)
--- a/modules/ocl/perf/perf_imgproc.cpp
+++ b/modules/ocl/perf/perf_imgproc.cpp
@@ -231,139 +231,6 @@ PERF_TEST_P(integralFixture, integral, OCL_TYPICAL_MAT_SIZES)
         OCL_PERF_ELSE
 }
 
-///////////// WarpAffine ////////////////////////
-
-typedef Size_MatType WarpAffineFixture;
-
-PERF_TEST_P(WarpAffineFixture, WarpAffine,
-            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
-                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4)))
-{
-    static const double coeffs[2][3] =
-    {
-        { cos(CV_PI / 6), -sin(CV_PI / 6), 100.0 },
-        { sin(CV_PI / 6), cos(CV_PI / 6), -100.0 }
-    };
-    Mat M(2, 3, CV_64F, (void *)coeffs);
-    const int interpolation = INTER_NEAREST;
-
-    const Size_MatType_t params = GetParam();
-    const Size srcSize = get<0>(params);
-    const int type = get<1>(params);
-
-    Mat src(srcSize, type), dst(srcSize, type);
-    declare.in(src, WARMUP_RNG).out(dst);
-
-    if (RUN_OCL_IMPL)
-    {
-        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
-
-        OCL_TEST_CYCLE() cv::ocl::warpAffine(oclSrc, oclDst, M, srcSize, interpolation);
-
-        oclDst.download(dst);
-
-        SANITY_CHECK(dst);
-    }
-    else if (RUN_PLAIN_IMPL)
-    {
-        TEST_CYCLE() cv::warpAffine(src, dst, M, srcSize, interpolation);
-
-        SANITY_CHECK(dst);
-    }
-    else
-        OCL_PERF_ELSE
-}
-
-///////////// WarpPerspective ////////////////////////
-
-typedef Size_MatType WarpPerspectiveFixture;
-
-PERF_TEST_P(WarpPerspectiveFixture, WarpPerspective,
-            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
-                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4)))
-{
-    static const double coeffs[3][3] =
-    {
-        {cos(CV_PI / 6), -sin(CV_PI / 6), 100.0},
-        {sin(CV_PI / 6), cos(CV_PI / 6), -100.0},
-        {0.0, 0.0, 1.0}
-    };
-    Mat M(3, 3, CV_64F, (void *)coeffs);
-    const int interpolation = INTER_LINEAR;
-
-    const Size_MatType_t params = GetParam();
-    const Size srcSize = get<0>(params);
-    const int type = get<1>(params);
-
-    Mat src(srcSize, type), dst(srcSize, type);
-    declare.in(src, WARMUP_RNG).out(dst)
-            .time(srcSize == OCL_SIZE_4000 ? 18 : srcSize == OCL_SIZE_2000 ? 5 : 2);
-
-    if (RUN_OCL_IMPL)
-    {
-        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
-
-        OCL_TEST_CYCLE() cv::ocl::warpPerspective(oclSrc, oclDst, M, srcSize, interpolation);
-
-        oclDst.download(dst);
-
-        SANITY_CHECK(dst);
-    }
-    else if (RUN_PLAIN_IMPL)
-    {
-        TEST_CYCLE() cv::warpPerspective(src, dst, M, srcSize, interpolation);
-
-        SANITY_CHECK(dst);
-    }
-    else
-        OCL_PERF_ELSE
-}
-
-///////////// resize ////////////////////////
-
-CV_ENUM(resizeInterType, INTER_NEAREST, INTER_LINEAR)
-
-typedef tuple<Size, MatType, resizeInterType, double> resizeParams;
-typedef TestBaseWithParam<resizeParams> resizeFixture;
-
-PERF_TEST_P(resizeFixture, resize,
-            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
-                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4),
-                               resizeInterType::all(),
-                               ::testing::Values(0.5, 2.0)))
-{
-    const resizeParams params = GetParam();
-    const Size srcSize = get<0>(params);
-    const int type = get<1>(params), interType = get<2>(params);
-    double scale = get<3>(params);
-
-    Mat src(srcSize, type), dst;
-    const Size dstSize(cvRound(srcSize.width * scale), cvRound(srcSize.height * scale));
-    dst.create(dstSize, type);
-    declare.in(src, WARMUP_RNG).out(dst);
-    if (interType == INTER_LINEAR && type == CV_8UC4 && OCL_SIZE_4000 == srcSize)
-        declare.time(11);
-
-    if (RUN_OCL_IMPL)
-    {
-        ocl::oclMat oclSrc(src), oclDst(dstSize, type);
-
-        OCL_TEST_CYCLE() cv::ocl::resize(oclSrc, oclDst, Size(), scale, scale, interType);
-
-        oclDst.download(dst);
-
-        SANITY_CHECK(dst, 1 + DBL_EPSILON);
-    }
-    else if (RUN_PLAIN_IMPL)
-    {
-        TEST_CYCLE() cv::resize(src, dst, Size(), scale, scale, interType);
-
-        SANITY_CHECK(dst, 1 + DBL_EPSILON);
-    }
-    else
-        OCL_PERF_ELSE
-}
-
 ///////////// threshold////////////////////////
 
 CV_ENUM(ThreshType, THRESH_BINARY, THRESH_TOZERO_INV)
@@ -727,67 +594,6 @@ PERF_TEST_P(meanShiftProcFixture, meanShiftProc,
         OCL_PERF_ELSE
 }
 
-///////////// remap////////////////////////
-
-CV_ENUM(RemapInterType, INTER_NEAREST, INTER_LINEAR)
-
-typedef tuple<Size, MatType, RemapInterType> remapParams;
-typedef TestBaseWithParam<remapParams> remapFixture;
-
-PERF_TEST_P(remapFixture, remap,
-            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
-                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4),
-                               RemapInterType::all()))
-{
-    const remapParams params = GetParam();
-    const Size srcSize = get<0>(params);
-    const int type = get<1>(params), interpolation = get<2>(params);
-
-    Mat src(srcSize, type), dst(srcSize, type);
-    declare.in(src, WARMUP_RNG).out(dst);
-
-    if (srcSize == OCL_SIZE_4000 && interpolation == INTER_LINEAR)
-        declare.time(9);
-
-    Mat xmap, ymap;
-    xmap.create(srcSize, CV_32FC1);
-    ymap.create(srcSize, CV_32FC1);
-
-    for (int i = 0; i < srcSize.height; ++i)
-    {
-        float * const xmap_row = xmap.ptr<float>(i);
-        float * const ymap_row = ymap.ptr<float>(i);
-
-        for (int j = 0; j < srcSize.width; ++j)
-        {
-            xmap_row[j] = (j - srcSize.width * 0.5f) * 0.75f + srcSize.width * 0.5f;
-            ymap_row[j] = (i - srcSize.height * 0.5f) * 0.75f + srcSize.height * 0.5f;
-        }
-    }
-
-    const int borderMode = BORDER_CONSTANT;
-
-    if (RUN_OCL_IMPL)
-    {
-        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
-        ocl::oclMat oclXMap(xmap), oclYMap(ymap);
-
-        OCL_TEST_CYCLE() cv::ocl::remap(oclSrc, oclDst, oclXMap, oclYMap, interpolation, borderMode);
-
-        oclDst.download(dst);
-
-        SANITY_CHECK(dst, 1 + DBL_EPSILON);
-    }
-    else if (RUN_PLAIN_IMPL)
-    {
-        TEST_CYCLE() cv::remap(src, dst, xmap, ymap, interpolation, borderMode);
-
-        SANITY_CHECK(dst, 1 + DBL_EPSILON);
-    }
-    else
-        OCL_PERF_ELSE
-}
-
 ///////////// CLAHE ////////////////////////
 
 typedef TestBaseWithParam<Size> CLAHEFixture;
@@ -872,58 +678,57 @@ PERF_TEST_P(columnSumFixture, columnSum, OCL_TYPICAL_MAT_SIZES)
 
 //////////////////////////////distanceToCenters////////////////////////////////////////////////
 
-CV_ENUM(DistType, NORM_L1, NORM_L2SQR);
+CV_ENUM(DistType, NORM_L1, NORM_L2SQR)
+
 typedef tuple<Size, DistType> distanceToCentersParameters;
 typedef TestBaseWithParam<distanceToCentersParameters> distanceToCentersFixture;
 
 static void distanceToCentersPerfTest(Mat& src, Mat& centers, Mat& dists, Mat& labels, int distType)
 {
     Mat batch_dists;
-    cv::batchDistance(src,centers,batch_dists, CV_32FC1, noArray(), distType);
+    cv::batchDistance(src, centers, batch_dists, CV_32FC1, noArray(), distType);
+
     std::vector<float> dists_v;
     std::vector<int> labels_v;
-    for(int i = 0; i<batch_dists.rows; i++)
+
+    for (int i = 0; i < batch_dists.rows; i++)
     {
         Mat r = batch_dists.row(i);
         double mVal;
         Point mLoc;
+
         minMaxLoc(r, &mVal, NULL, &mLoc, NULL);
-        dists_v.push_back((float)mVal);
+        dists_v.push_back(static_cast<float>(mVal));
         labels_v.push_back(mLoc.x);
     }
-    Mat temp_dists(dists_v);
-    Mat temp_labels(labels_v);
-    temp_dists.reshape(1,1).copyTo(dists);
-    temp_labels.reshape(1,1).copyTo(labels);
+
+    Mat(dists_v).copyTo(dists);
+    Mat(labels_v).copyTo(labels);
 }
 
 PERF_TEST_P(distanceToCentersFixture, distanceToCenters, ::testing::Combine(::testing::Values(cv::Size(256,256), cv::Size(512,512)), DistType::all()) )
 {
     Size size = get<0>(GetParam());
     int distType = get<1>(GetParam());
-    Mat src(size, CV_32FC1);
-    Mat centers(size, CV_32FC1);
-    Mat dists(cv::Size(src.rows,1), CV_32FC1);
-    Mat labels(cv::Size(src.rows,1), CV_32SC1);
+
+    Mat src(size, CV_32FC1), centers(size, CV_32FC1);
+    Mat dists(src.rows, 1, CV_32FC1), labels(src.rows, 1, CV_32SC1);
+
     declare.in(src, centers, WARMUP_RNG).out(dists, labels);
+
     if (RUN_OCL_IMPL)
     {
-        ocl::oclMat ocl_src(src);
-        ocl::oclMat ocl_centers(centers);
-        ocl::oclMat ocl_dists(dists);
-        ocl::oclMat ocl_labels(labels);
+        ocl::oclMat ocl_src(src), ocl_centers(centers);
 
-        OCL_TEST_CYCLE() ocl::distanceToCenters(ocl_dists,ocl_labels,ocl_src, ocl_centers, distType);
-
-        ocl_dists.download(dists);
-        ocl_labels.download(labels);
+        OCL_TEST_CYCLE() ocl::distanceToCenters(ocl_src, ocl_centers, dists, labels, distType);
 
         SANITY_CHECK(dists, 1e-6, ERROR_RELATIVE);
         SANITY_CHECK(labels);
     }
     else if (RUN_PLAIN_IMPL)
     {
-        TEST_CYCLE() distanceToCentersPerfTest(src,centers,dists,labels,distType);
+        TEST_CYCLE() distanceToCentersPerfTest(src, centers, dists, labels, distType);
+
         SANITY_CHECK(dists, 1e-6, ERROR_RELATIVE);
         SANITY_CHECK(labels);
     }

diff --git a/modules/ocl/perf/perf_imgwarp.cpp b/modules/ocl/perf/perf_imgwarp.cpp
new file mode 100644 (file)
index 0000000..e768d66
--- /dev/null
+++ b/modules/ocl/perf/perf_imgwarp.cpp
@@ -0,0 +1,364 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2010-2012, Multicoreware, Inc., all rights reserved.
+// Copyright (C) 2010-2012, Advanced Micro Devices, Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// @Authors
+//    Fangfang Bai, fangfang@multicorewareinc.com
+//    Jin Ma,       jin@multicorewareinc.com
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors as is and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+#include "perf_precomp.hpp"
+
+using namespace perf;
+using std::tr1::tuple;
+using std::tr1::get;
+
+///////////// WarpAffine ////////////////////////
+
+typedef Size_MatType WarpAffineFixture;
+
+PERF_TEST_P(WarpAffineFixture, WarpAffine,
+            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
+                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4)))
+{
+    static const double coeffs[2][3] =
+    {
+        { cos(CV_PI / 6), -sin(CV_PI / 6), 100.0 },
+        { sin(CV_PI / 6), cos(CV_PI / 6), -100.0 }
+    };
+    Mat M(2, 3, CV_64F, (void *)coeffs);
+    const int interpolation = INTER_NEAREST;
+
+    const Size_MatType_t params = GetParam();
+    const Size srcSize = get<0>(params);
+    const int type = get<1>(params);
+
+    Mat src(srcSize, type), dst(srcSize, type);
+    declare.in(src, WARMUP_RNG).out(dst);
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
+
+        OCL_TEST_CYCLE() cv::ocl::warpAffine(oclSrc, oclDst, M, srcSize, interpolation);
+
+        oclDst.download(dst);
+
+        SANITY_CHECK(dst);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() cv::warpAffine(src, dst, M, srcSize, interpolation);
+
+        SANITY_CHECK(dst);
+    }
+    else
+        OCL_PERF_ELSE
+}
+
+///////////// WarpPerspective ////////////////////////
+
+typedef Size_MatType WarpPerspectiveFixture;
+
+PERF_TEST_P(WarpPerspectiveFixture, WarpPerspective,
+            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
+                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4)))
+{
+    static const double coeffs[3][3] =
+    {
+        {cos(CV_PI / 6), -sin(CV_PI / 6), 100.0},
+        {sin(CV_PI / 6), cos(CV_PI / 6), -100.0},
+        {0.0, 0.0, 1.0}
+    };
+    Mat M(3, 3, CV_64F, (void *)coeffs);
+    const int interpolation = INTER_LINEAR;
+
+    const Size_MatType_t params = GetParam();
+    const Size srcSize = get<0>(params);
+    const int type = get<1>(params);
+
+    Mat src(srcSize, type), dst(srcSize, type);
+    declare.in(src, WARMUP_RNG).out(dst)
+            .time(srcSize == OCL_SIZE_4000 ? 18 : srcSize == OCL_SIZE_2000 ? 5 : 2);
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
+
+        OCL_TEST_CYCLE() cv::ocl::warpPerspective(oclSrc, oclDst, M, srcSize, interpolation);
+
+        oclDst.download(dst);
+
+        SANITY_CHECK(dst);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() cv::warpPerspective(src, dst, M, srcSize, interpolation);
+
+        SANITY_CHECK(dst);
+    }
+    else
+        OCL_PERF_ELSE
+}
+
+///////////// resize ////////////////////////
+
+CV_ENUM(resizeInterType, INTER_NEAREST, INTER_LINEAR)
+
+typedef tuple<Size, MatType, resizeInterType, double> resizeParams;
+typedef TestBaseWithParam<resizeParams> resizeFixture;
+
+PERF_TEST_P(resizeFixture, resize,
+            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
+                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4),
+                               resizeInterType::all(),
+                               ::testing::Values(0.5, 2.0)))
+{
+    const resizeParams params = GetParam();
+    const Size srcSize = get<0>(params);
+    const int type = get<1>(params), interType = get<2>(params);
+    double scale = get<3>(params);
+    const Size dstSize(cvRound(srcSize.width * scale), cvRound(srcSize.height * scale));
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+    checkDeviceMaxMemoryAllocSize(dstSize, type);
+
+    Mat src(srcSize, type), dst;
+    dst.create(dstSize, type);
+    declare.in(src, WARMUP_RNG).out(dst);
+    if (interType == INTER_LINEAR && type == CV_8UC4 && OCL_SIZE_4000 == srcSize)
+        declare.time(11);
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclSrc(src), oclDst(dstSize, type);
+
+        OCL_TEST_CYCLE() cv::ocl::resize(oclSrc, oclDst, Size(), scale, scale, interType);
+
+        oclDst.download(dst);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() cv::resize(src, dst, Size(), scale, scale, interType);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else
+        OCL_PERF_ELSE
+}
+
+typedef tuple<Size, MatType, double> resizeAreaParams;
+typedef TestBaseWithParam<resizeAreaParams> resizeAreaFixture;
+
+PERF_TEST_P(resizeAreaFixture, resize,
+            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
+                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4),
+                               ::testing::Values(0.3, 0.5, 0.6)))
+{
+    const resizeAreaParams params = GetParam();
+    const Size srcSize = get<0>(params);
+    const int type = get<1>(params);
+    double scale = get<2>(params);
+    const Size dstSize(cvRound(srcSize.width * scale), cvRound(srcSize.height * scale));
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+
+    Mat src(srcSize, type), dst;
+    dst.create(dstSize, type);
+    declare.in(src, WARMUP_RNG).out(dst);
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclSrc(src), oclDst(dstSize, type);
+
+        OCL_TEST_CYCLE() cv::ocl::resize(oclSrc, oclDst, Size(), scale, scale, cv::INTER_AREA);
+
+        oclDst.download(dst);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() cv::resize(src, dst, Size(), scale, scale, cv::INTER_AREA);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else
+        OCL_PERF_ELSE
+}
+
+///////////// remap////////////////////////
+
+CV_ENUM(RemapInterType, INTER_NEAREST, INTER_LINEAR)
+
+typedef tuple<Size, MatType, RemapInterType> remapParams;
+typedef TestBaseWithParam<remapParams> remapFixture;
+
+PERF_TEST_P(remapFixture, remap,
+            ::testing::Combine(OCL_TYPICAL_MAT_SIZES,
+                               OCL_PERF_ENUM(CV_8UC1, CV_8UC4),
+                               RemapInterType::all()))
+{
+    const remapParams params = GetParam();
+    const Size srcSize = get<0>(params);
+    const int type = get<1>(params), interpolation = get<2>(params);
+
+    Mat src(srcSize, type), dst(srcSize, type);
+    declare.in(src, WARMUP_RNG).out(dst);
+
+    if (srcSize == OCL_SIZE_4000 && interpolation == INTER_LINEAR)
+        declare.time(9);
+
+    Mat xmap, ymap;
+    xmap.create(srcSize, CV_32FC1);
+    ymap.create(srcSize, CV_32FC1);
+
+    for (int i = 0; i < srcSize.height; ++i)
+    {
+        float * const xmap_row = xmap.ptr<float>(i);
+        float * const ymap_row = ymap.ptr<float>(i);
+
+        for (int j = 0; j < srcSize.width; ++j)
+        {
+            xmap_row[j] = (j - srcSize.width * 0.5f) * 0.75f + srcSize.width * 0.5f;
+            ymap_row[j] = (i - srcSize.height * 0.5f) * 0.75f + srcSize.height * 0.5f;
+        }
+    }
+
+    const int borderMode = BORDER_CONSTANT;
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclSrc(src), oclDst(srcSize, type);
+        ocl::oclMat oclXMap(xmap), oclYMap(ymap);
+
+        OCL_TEST_CYCLE() cv::ocl::remap(oclSrc, oclDst, oclXMap, oclYMap, interpolation, borderMode);
+
+        oclDst.download(dst);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() cv::remap(src, dst, xmap, ymap, interpolation, borderMode);
+
+        SANITY_CHECK(dst, 1 + DBL_EPSILON);
+    }
+    else
+        OCL_PERF_ELSE
+}
+
+
+///////////// buildWarpPerspectiveMaps ////////////////////////
+
+static void buildWarpPerspectiveMaps(const Mat &M, bool inverse, Size dsize, Mat &xmap, Mat &ymap)
+{
+    CV_Assert(M.rows == 3 && M.cols == 3);
+    CV_Assert(dsize.area() > 0);
+
+    xmap.create(dsize, CV_32FC1);
+    ymap.create(dsize, CV_32FC1);
+
+    float coeffs[3 * 3];
+    Mat coeffsMat(3, 3, CV_32F, (void *)coeffs);
+
+    if (inverse)
+        M.convertTo(coeffsMat, coeffsMat.type());
+    else
+    {
+        cv::Mat iM;
+        invert(M, iM);
+        iM.convertTo(coeffsMat, coeffsMat.type());
+    }
+
+    for (int y = 0; y < dsize.height; ++y)
+    {
+        float * const xmap_ptr = xmap.ptr<float>(y);
+        float * const ymap_ptr = ymap.ptr<float>(y);
+
+        for (int x = 0; x < dsize.width; ++x)
+        {
+            float coeff = 1.0f / (x * coeffs[6] + y * coeffs[7] + coeffs[8]);
+            xmap_ptr[x] = (x * coeffs[0] + y * coeffs[1] + coeffs[2]) * coeff;
+            ymap_ptr[x] = (x * coeffs[3] + y * coeffs[4] + coeffs[5]) * coeff;
+        }
+    }
+}
+
+typedef TestBaseWithParam<Size> buildWarpPerspectiveMapsFixture;
+
+PERF_TEST_P(buildWarpPerspectiveMapsFixture, Inverse, OCL_TYPICAL_MAT_SIZES)
+{
+    static const double coeffs[3][3] =
+    {
+        {cos(CV_PI / 6), -sin(CV_PI / 6), 100.0},
+        {sin(CV_PI / 6), cos(CV_PI / 6), -100.0},
+        {0.0, 0.0, 1.0}
+    };
+    Mat M(3, 3, CV_64F, (void *)coeffs);
+    const Size dsize = GetParam();
+    const double eps = 5e-4;
+
+    Mat xmap(dsize, CV_32FC1), ymap(dsize, CV_32FC1);
+    declare.in(M).out(xmap, ymap);
+
+    if (RUN_OCL_IMPL)
+    {
+        ocl::oclMat oclXMap(dsize, CV_32FC1), oclYMap(dsize, CV_32FC1);
+
+        OCL_TEST_CYCLE() cv::ocl::buildWarpPerspectiveMaps(M, true, dsize, oclXMap, oclYMap);
+
+        oclXMap.download(xmap);
+        oclYMap.download(ymap);
+
+        SANITY_CHECK(xmap, eps);
+        SANITY_CHECK(ymap, eps);
+    }
+    else if (RUN_PLAIN_IMPL)
+    {
+        TEST_CYCLE() buildWarpPerspectiveMaps(M, true, dsize, xmap, ymap);
+
+        SANITY_CHECK(xmap, eps);
+        SANITY_CHECK(ymap, eps);
+    }
+    else
+        OCL_PERF_ELSE
+}

diff --git a/modules/ocl/perf/perf_match_template.cpp b/modules/ocl/perf/perf_match_template.cpp
index d6f7fe0..3ee038a 100644 (file)
--- a/modules/ocl/perf/perf_match_template.cpp
+++ b/modules/ocl/perf/perf_match_template.cpp
@@ -108,13 +108,13 @@ PERF_TEST_P(CV_TM_CCORR_NORMEDFixture, matchTemplate, OCL_TYPICAL_MAT_SIZES)
 
         oclDst.download(dst);
 
-        SANITY_CHECK(dst, 2e-2);
+        SANITY_CHECK(dst, 3e-2);
     }
     else if (RUN_PLAIN_IMPL)
     {
         TEST_CYCLE() cv::matchTemplate(src, templ, dst, CV_TM_CCORR_NORMED);
 
-        SANITY_CHECK(dst, 2e-2);
+        SANITY_CHECK(dst, 3e-2);
     }
     else
         OCL_PERF_ELSE

diff --git a/modules/ocl/perf/perf_matrix_operation.cpp b/modules/ocl/perf/perf_matrix_operation.cpp
index f2baa7f..5ca322e 100644 (file)
--- a/modules/ocl/perf/perf_matrix_operation.cpp
+++ b/modules/ocl/perf/perf_matrix_operation.cpp
@@ -63,6 +63,10 @@ PERF_TEST_P(ConvertToFixture, ConvertTo,
 
     Mat src(srcSize, type), dst;
     const int dstType = CV_MAKE_TYPE(CV_32F, src.channels());
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+    checkDeviceMaxMemoryAllocSize(srcSize, dstType);
+
     dst.create(srcSize, dstType);
     declare.in(src, WARMUP_RNG).out(dst);
 

diff --git a/modules/ocl/perf/perf_precomp.hpp b/modules/ocl/perf/perf_precomp.hpp
index 2d9639a..a8663df 100644 (file)
--- a/modules/ocl/perf/perf_precomp.hpp
+++ b/modules/ocl/perf/perf_precomp.hpp
@@ -103,8 +103,30 @@ using namespace cv;
             CV_TEST_FAIL_NO_IMPL();
 #endif
 
-#define OCL_TEST_CYCLE_N(n) for(declare.iterations(n); startTimer(), next(); ocl::finish(), stopTimer())
-#define OCL_TEST_CYCLE() for(; startTimer(), next(); ocl::finish(), stopTimer())
-#define OCL_TEST_CYCLE_MULTIRUN(runsNum) for(declare.runs(runsNum); startTimer(), next(); stopTimer()) for(int r = 0; r < runsNum; ocl::finish(), ++r)
+#define OCL_TEST_CYCLE_N(n) for(declare.iterations(n); startTimer(), next(); cv::ocl::finish(), stopTimer())
+#define OCL_TEST_CYCLE() for(; startTimer(), next(); cv::ocl::finish(), stopTimer())
+#define OCL_TEST_CYCLE_MULTIRUN(runsNum) for(declare.runs(runsNum); startTimer(), next(); stopTimer()) for(int r = 0; r < runsNum; cv::ocl::finish(), ++r)
+
+namespace cvtest {
+namespace ocl {
+inline void checkDeviceMaxMemoryAllocSize(const Size& size, int type, int factor = 1)
+{
+    assert(factor > 0);
+    if (!(IMPL_OCL == perf::TestBase::getSelectedImpl()))
+        return; // OpenCL devices are not used
+    int cn = CV_MAT_CN(type);
+    int cn_ocl = cn == 3 ? 4 : cn;
+    int type_ocl = CV_MAKE_TYPE(CV_MAT_DEPTH(type), cn_ocl);
+    size_t memSize = size.area() * CV_ELEM_SIZE(type_ocl);
+    const cv::ocl::DeviceInfo& devInfo = cv::ocl::Context::getContext()->getDeviceInfo();
+    if (memSize * factor >= devInfo.maxMemAllocSize)
+    {
+        throw perf::TestBase::PerfSkipTestException();
+    }
+}
+} // namespace cvtest::ocl
+} // namespace cvtest
+
+using namespace cvtest::ocl;
 
 #endif

diff --git a/modules/ocl/perf/perf_pyramid.cpp b/modules/ocl/perf/perf_pyramid.cpp
index c799853..820dd60 100644 (file)
--- a/modules/ocl/perf/perf_pyramid.cpp
+++ b/modules/ocl/perf/perf_pyramid.cpp
@@ -60,9 +60,12 @@ PERF_TEST_P(pyrDownFixture, pyrDown,
     const Size_MatType_t params = GetParam();
     const Size srcSize = get<0>(params);
     const int type = get<1>(params);
+    Size dstSize((srcSize.height + 1) >> 1, (srcSize.width + 1) >> 1);
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+    checkDeviceMaxMemoryAllocSize(dstSize, type);
 
     Mat src(srcSize, type), dst;
-    Size dstSize((srcSize.height + 1) >> 1, (srcSize.width + 1) >> 1);
     dst.create(dstSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 
@@ -97,9 +100,12 @@ PERF_TEST_P(pyrUpFixture, pyrUp,
     const Size_MatType_t params = GetParam();
     const Size srcSize = get<0>(params);
     const int type = get<1>(params);
+    Size dstSize(srcSize.height << 1, srcSize.width << 1);
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
+    checkDeviceMaxMemoryAllocSize(dstSize, type);
 
     Mat src(srcSize, type), dst;
-    Size dstSize(srcSize.height << 1, srcSize.width << 1);
     dst.create(dstSize, type);
     declare.in(src, WARMUP_RNG).out(dst);
 

diff --git a/modules/ocl/perf/perf_split_merge.cpp b/modules/ocl/perf/perf_split_merge.cpp
index f2f7c41..ecfc49e 100644 (file)
--- a/modules/ocl/perf/perf_split_merge.cpp
+++ b/modules/ocl/perf/perf_split_merge.cpp
@@ -60,8 +60,10 @@ PERF_TEST_P(MergeFixture, Merge,
     const Size_MatType_t params = GetParam();
     const Size srcSize = get<0>(params);
     const int depth = get<1>(params), channels = 3;
-
     const int dstType = CV_MAKE_TYPE(depth, channels);
+
+    checkDeviceMaxMemoryAllocSize(srcSize, dstType);
+
     Mat dst(srcSize, dstType);
     vector<Mat> src(channels);
     for (vector<Mat>::iterator i = src.begin(), end = src.end(); i != end; ++i)
@@ -105,8 +107,11 @@ PERF_TEST_P(SplitFixture, Split,
     const Size_MatType_t params = GetParam();
     const Size srcSize = get<0>(params);
     const int depth = get<1>(params), channels = 3;
+    const int type = CV_MAKE_TYPE(depth, channels);
+
+    checkDeviceMaxMemoryAllocSize(srcSize, type);
 
-    Mat src(srcSize, CV_MAKE_TYPE(depth, channels));
+    Mat src(srcSize, type);
     declare.in(src, WARMUP_RNG);
 
     if (RUN_OCL_IMPL)

diff --git a/modules/ocl/src/arithm.cpp b/modules/ocl/src/arithm.cpp
index f8a0690..8a2390d 100644 (file)
--- a/modules/ocl/src/arithm.cpp
+++ b/modules/ocl/src/arithm.cpp
@@ -394,12 +394,16 @@ Scalar cv::ocl::sum(const oclMat &src)
 
 Scalar cv::ocl::absSum(const oclMat &src)
 {
-    if (!src.clCxt->supportsFeature(FEATURE_CL_DOUBLE) && src.depth() == CV_64F)
+    int sdepth = src.depth();
+    if (!src.clCxt->supportsFeature(FEATURE_CL_DOUBLE) && sdepth == CV_64F)
     {
         CV_Error(CV_OpenCLDoubleNotSupported, "Selected device doesn't support double");
         return cv::Scalar::all(0);
     }
 
+    if (sdepth == CV_8U || sdepth == CV_16U)
+        return sum(src);
+
     static sumFunc functab[3] =
     {
         arithmetic_sum<int>,
@@ -407,7 +411,7 @@ Scalar cv::ocl::absSum(const oclMat &src)
         arithmetic_sum<double>
     };
 
-    int ddepth = std::max(src.depth(), CV_32S);
+    int ddepth = std::max(sdepth, CV_32S);
     sumFunc func = functab[ddepth - CV_32S];
     return func(src, ABS_SUM, ddepth);
 }
@@ -867,30 +871,32 @@ void cv::ocl::log(const oclMat &src, oclMat &dst)
 
 static void arithmetic_magnitude_phase_run(const oclMat &src1, const oclMat &src2, oclMat &dst, string kernelName)
 {
-    int channels = dst.oclchannels();
     int depth = dst.depth();
 
-    size_t vector_length = 1;
-    int offset_cols = ((dst.offset % dst.step) / dst.elemSize1()) & (vector_length - 1);
-    int cols = divUp(dst.cols * channels + offset_cols, vector_length);
-
     size_t localThreads[3]  = { 64, 4, 1 };
-    size_t globalThreads[3] = { cols, dst.rows, 1 };
+    size_t globalThreads[3] = { dst.cols, dst.rows, 1 };
+
+    int src1_step = src1.step / src1.elemSize(), src1_offset = src1.offset / src1.elemSize();
+    int src2_step = src2.step / src2.elemSize(), src2_offset = src2.offset / src2.elemSize();
+    int dst_step = dst.step / dst.elemSize(), dst_offset = dst.offset / dst.elemSize();
 
     vector<pair<size_t , const void *> > args;
     args.push_back( make_pair( sizeof(cl_mem), (void *)&src1.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src1_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src1_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&src2.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_offset ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&dst.rows ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&cols ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.cols ));
 
-    openCLExecuteKernel(src1.clCxt, &arithm_magnitude, kernelName, globalThreads, localThreads, args, -1, depth);
+    const char * const channelMap[] = { "", "", "2", "4", "4" };
+    std::string buildOptions = format("-D T=%s%s", depth == CV_32F ? "float" : "double", channelMap[dst.channels()]);
+
+    openCLExecuteKernel(src1.clCxt, &arithm_magnitude, kernelName, globalThreads, localThreads, args, -1, -1, buildOptions.c_str());
 }
 
 void cv::ocl::magnitude(const oclMat &src1, const oclMat &src2, oclMat &dst)
@@ -964,25 +970,29 @@ static void arithmetic_cartToPolar_run(const oclMat &src1, const oclMat &src2, o
     size_t localThreads[3]  = { 64, 4, 1 };
     size_t globalThreads[3] = { cols, src1.rows, 1 };
 
-    int tmp = angleInDegrees ? 1 : 0;
+    int src1_step = src1.step / src1.elemSize1(), src1_offset = src1.offset / src1.elemSize1();
+    int src2_step = src2.step / src2.elemSize1(), src2_offset = src2.offset / src2.elemSize1();
+    int dst_mag_step = dst_mag.step / dst_mag.elemSize1(), dst_mag_offset = dst_mag.offset / dst_mag.elemSize1();
+    int dst_cart_step = dst_cart.step / dst_cart.elemSize1(), dst_cart_offset = dst_cart.offset / dst_cart.elemSize1();
+
     vector<pair<size_t , const void *> > args;
     args.push_back( make_pair( sizeof(cl_mem), (void *)&src1.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src1_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src1_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&src2.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst_mag.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_mag.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_mag.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_mag_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_mag_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst_cart.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_cart.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_cart.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_cart_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_cart_offset ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&src1.rows ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&cols ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&tmp ));
 
-    openCLExecuteKernel(src1.clCxt, &arithm_cartToPolar, kernelName, globalThreads, localThreads, args, -1, depth);
+    openCLExecuteKernel(src1.clCxt, &arithm_cartToPolar, kernelName, globalThreads, localThreads, args,
+                        -1, depth, angleInDegrees ? "-D DEGREE" : "-D RADIAN");
 }
 
 void cv::ocl::cartToPolar(const oclMat &x, const oclMat &y, oclMat &mag, oclMat &angle, bool angleInDegrees)
@@ -1008,37 +1018,38 @@ void cv::ocl::cartToPolar(const oclMat &x, const oclMat &y, oclMat &mag, oclMat
 static void arithmetic_ptc_run(const oclMat &src1, const oclMat &src2, oclMat &dst1, oclMat &dst2, bool angleInDegrees,
                         string kernelName)
 {
-    int channels = src2.oclchannels();
-    int depth = src2.depth();
-
-    int cols = src2.cols * channels;
-    int rows = src2.rows;
+    int channels = src2.oclchannels(), depth = src2.depth();
+    int cols = src2.cols * channels, rows = src2.rows;
 
     size_t localThreads[3]  = { 64, 4, 1 };
     size_t globalThreads[3] = { cols, rows, 1 };
 
-    int tmp = angleInDegrees ? 1 : 0;
+    int src1_step = src1.step / src1.elemSize1(), src1_offset = src1.offset / src1.elemSize1();
+    int src2_step = src2.step / src2.elemSize1(), src2_offset = src2.offset / src2.elemSize1();
+    int dst1_step = dst1.step / dst1.elemSize1(), dst1_offset = dst1.offset / dst1.elemSize1();
+    int dst2_step = dst2.step / dst2.elemSize1(), dst2_offset = dst2.offset / dst2.elemSize1();
+
     vector<pair<size_t , const void *> > args;
     if (src1.data)
     {
         args.push_back( make_pair( sizeof(cl_mem), (void *)&src1.data ));
-        args.push_back( make_pair( sizeof(cl_int), (void *)&src1.step ));
-        args.push_back( make_pair( sizeof(cl_int), (void *)&src1.offset ));
+        args.push_back( make_pair( sizeof(cl_int), (void *)&src1_step ));
+        args.push_back( make_pair( sizeof(cl_int), (void *)&src1_offset ));
     }
     args.push_back( make_pair( sizeof(cl_mem), (void *)&src2.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src2.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src2_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst1.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst1.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst1.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst1_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst1_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst2.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst2.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst2.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst2_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst2_offset ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&rows ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&cols ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&tmp ));
 
-    openCLExecuteKernel(src1.clCxt, &arithm_polarToCart, kernelName, globalThreads, localThreads, args, -1, depth);
+    openCLExecuteKernel(src1.clCxt, &arithm_polarToCart, kernelName, globalThreads, localThreads,
+                        args, -1, depth, angleInDegrees ? "-D DEGREE" : "-D RADIAN");
 }
 
 void cv::ocl::polarToCart(const oclMat &magnitude, const oclMat &angle, oclMat &x, oclMat &y, bool angleInDegrees)
@@ -1623,38 +1634,38 @@ void cv::ocl::addWeighted(const oclMat &src1, double alpha, const oclMat &src2,
 /////////////////////////////////// Pow //////////////////////////////////////
 //////////////////////////////////////////////////////////////////////////////
 
-static void arithmetic_pow_run(const oclMat &src1, double p, oclMat &dst, string kernelName, const cv::ocl::ProgramEntry* source)
+static void arithmetic_pow_run(const oclMat &src, double p, oclMat &dst, string kernelName, const cv::ocl::ProgramEntry* source)
 {
     int channels = dst.oclchannels();
     int depth = dst.depth();
 
-    size_t vector_length = 1;
-    int offset_cols = ((dst.offset % dst.step) / dst.elemSize1()) & (vector_length - 1);
-    int cols = divUp(dst.cols * channels + offset_cols, vector_length);
-    int rows = dst.rows;
-
     size_t localThreads[3]  = { 64, 4, 1 };
-    size_t globalThreads[3] = { cols, rows, 1 };
+    size_t globalThreads[3] = { dst.cols, dst.rows, 1 };
+
+    const char * const typeStr = depth == CV_32F ? "float" : "double";
+    const char * const channelMap[] = { "", "", "2", "4", "4" };
+    std::string buildOptions = format("-D VT=%s%s -D T=%s", typeStr, channelMap[channels], typeStr);
+
+    int src_step = src.step / src.elemSize(), src_offset = src.offset / src.elemSize();
+    int dst_step = dst.step / dst.elemSize(), dst_offset = dst.offset / dst.elemSize();
 
-    int dst_step1 = dst.cols * dst.elemSize();
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&src1.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&src1.offset ));
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&src.data ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&dst.data ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.step ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.offset ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_step ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_offset ));
     args.push_back( make_pair( sizeof(cl_int), (void *)&dst.rows ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&cols ));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&dst_step1 ));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&dst.cols ));
 
     float pf = static_cast<float>(p);
-    if (!src1.clCxt->supportsFeature(FEATURE_CL_DOUBLE))
+    if(src.depth() == CV_32F)
         args.push_back( make_pair( sizeof(cl_float), (void *)&pf ));
     else
         args.push_back( make_pair( sizeof(cl_double), (void *)&p ));
 
-    openCLExecuteKernel(src1.clCxt, source, kernelName, globalThreads, localThreads, args, -1, depth);
+    openCLExecuteKernel(src.clCxt, source, kernelName, globalThreads, localThreads, args, -1, -1, buildOptions.c_str());
 }
 
 void cv::ocl::pow(const oclMat &x, double p, oclMat &y)

diff --git a/modules/ocl/src/build_warps.cpp b/modules/ocl/src/build_warps.cpp
index dc9ab66..40c082b 100644 (file)
--- a/modules/ocl/src/build_warps.cpp
+++ b/modules/ocl/src/build_warps.cpp
@@ -53,7 +53,7 @@ using namespace cv::ocl;
 // buildWarpPlaneMaps
 
 void cv::ocl::buildWarpPlaneMaps(Size /*src_size*/, Rect dst_roi, const Mat &K, const Mat &R, const Mat &T,
-                                 float scale, oclMat &map_x, oclMat &map_y)
+                                 float scale, oclMat &xmap, oclMat &ymap)
 {
     CV_Assert(K.size() == Size(3, 3) && K.type() == CV_32F);
     CV_Assert(R.size() == Size(3, 3) && R.type() == CV_32F);
@@ -68,37 +68,40 @@ void cv::ocl::buildWarpPlaneMaps(Size /*src_size*/, Rect dst_roi, const Mat &K,
 
     oclMat KRT_oclMat(KRT_mat);
     // transfer K_Rinv and T into a single cl_mem
-    map_x.create(dst_roi.size(), CV_32F);
-    map_y.create(dst_roi.size(), CV_32F);
+    xmap.create(dst_roi.size(), CV_32F);
+    ymap.create(dst_roi.size(), CV_32F);
 
     int tl_u = dst_roi.tl().x;
     int tl_v = dst_roi.tl().y;
 
-    Context *clCxt = Context::getContext();
-    string kernelName = "buildWarpPlaneMaps";
-    vector< pair<size_t, const void *> > args;
+    int xmap_step = xmap.step / xmap.elemSize(), xmap_offset = xmap.offset / xmap.elemSize();
+    int ymap_step = ymap.step / ymap.elemSize(), ymap_offset = ymap.offset / ymap.elemSize();
 
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_x.data));
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_y.data));
+    vector< pair<size_t, const void *> > args;
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&xmap.data));
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&ymap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&KRT_mat.data));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_u));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_v));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.cols));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.rows));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.step));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_y.step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.cols));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.rows));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_offset));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_offset));
     args.push_back( make_pair( sizeof(cl_float), (void *)&scale));
 
-    size_t globalThreads[3] = {map_x.cols, map_x.rows, 1};
-    size_t localThreads[3]  = {32, 8, 1};
-    openCLExecuteKernel(clCxt, &build_warps, kernelName, globalThreads, localThreads, args, -1, -1);
+    size_t globalThreads[3] = { xmap.cols, xmap.rows, 1 };
+    size_t localThreads[3]  = { 32, 8, 1 };
+
+    openCLExecuteKernel(Context::getContext(), &build_warps, "buildWarpPlaneMaps", globalThreads, localThreads, args, -1, -1);
 }
 
 //////////////////////////////////////////////////////////////////////////////
 // buildWarpCylyndricalMaps
 
 void cv::ocl::buildWarpCylindricalMaps(Size /*src_size*/, Rect dst_roi, const Mat &K, const Mat &R, float scale,
-                                       oclMat &map_x, oclMat &map_y)
+                                       oclMat &xmap, oclMat &ymap)
 {
     CV_Assert(K.size() == Size(3, 3) && K.type() == CV_32F);
     CV_Assert(R.size() == Size(3, 3) && R.type() == CV_32F);
@@ -108,36 +111,40 @@ void cv::ocl::buildWarpCylindricalMaps(Size /*src_size*/, Rect dst_roi, const Ma
 
     oclMat KR_oclMat(K_Rinv.reshape(1, 1));
 
-    map_x.create(dst_roi.size(), CV_32F);
-    map_y.create(dst_roi.size(), CV_32F);
+    xmap.create(dst_roi.size(), CV_32F);
+    ymap.create(dst_roi.size(), CV_32F);
 
     int tl_u = dst_roi.tl().x;
     int tl_v = dst_roi.tl().y;
 
-    Context *clCxt = Context::getContext();
-    string kernelName = "buildWarpCylindricalMaps";
-    vector< pair<size_t, const void *> > args;
+    int xmap_step = xmap.step / xmap.elemSize(), xmap_offset = xmap.offset / xmap.elemSize();
+    int ymap_step = ymap.step / ymap.elemSize(), ymap_offset = ymap.offset / ymap.elemSize();
 
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_x.data));
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_y.data));
+    vector< pair<size_t, const void *> > args;
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&xmap.data));
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&ymap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&KR_oclMat.data));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_u));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_v));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.cols));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.rows));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.step));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_y.step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.cols));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.rows));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_offset));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_offset));
     args.push_back( make_pair( sizeof(cl_float), (void *)&scale));
 
-    size_t globalThreads[3] = {map_x.cols, map_x.rows, 1};
-    size_t localThreads[3]  = {32, 8, 1};
-    openCLExecuteKernel(clCxt, &build_warps, kernelName, globalThreads, localThreads, args, -1, -1);
+    size_t globalThreads[3] = { xmap.cols, xmap.rows, 1 };
+    size_t localThreads[3]  = { 32, 8, 1 };
+
+    openCLExecuteKernel(Context::getContext(), &build_warps, "buildWarpCylindricalMaps", globalThreads, localThreads, args, -1, -1);
 }
 
 //////////////////////////////////////////////////////////////////////////////
 // buildWarpSphericalMaps
+
 void cv::ocl::buildWarpSphericalMaps(Size /*src_size*/, Rect dst_roi, const Mat &K, const Mat &R, float scale,
-                                     oclMat &map_x, oclMat &map_y)
+                                     oclMat &xmap, oclMat &ymap)
 {
     CV_Assert(K.size() == Size(3, 3) && K.type() == CV_32F);
     CV_Assert(R.size() == Size(3, 3) && R.type() == CV_32F);
@@ -147,37 +154,41 @@ void cv::ocl::buildWarpSphericalMaps(Size /*src_size*/, Rect dst_roi, const Mat
 
     oclMat KR_oclMat(K_Rinv.reshape(1, 1));
     // transfer K_Rinv, R_Kinv into a single cl_mem
-    map_x.create(dst_roi.size(), CV_32F);
-    map_y.create(dst_roi.size(), CV_32F);
+    xmap.create(dst_roi.size(), CV_32F);
+    ymap.create(dst_roi.size(), CV_32F);
 
     int tl_u = dst_roi.tl().x;
     int tl_v = dst_roi.tl().y;
 
-    Context *clCxt = Context::getContext();
-    string kernelName = "buildWarpSphericalMaps";
-    vector< pair<size_t, const void *> > args;
+    int xmap_step = xmap.step / xmap.elemSize(), xmap_offset = xmap.offset / xmap.elemSize();
+    int ymap_step = ymap.step / ymap.elemSize(), ymap_offset = ymap.offset / ymap.elemSize();
 
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_x.data));
-    args.push_back( make_pair( sizeof(cl_mem), (void *)&map_y.data));
+    vector< pair<size_t, const void *> > args;
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&xmap.data));
+    args.push_back( make_pair( sizeof(cl_mem), (void *)&ymap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&KR_oclMat.data));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_u));
     args.push_back( make_pair( sizeof(cl_int), (void *)&tl_v));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.cols));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.rows));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_x.step));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&map_y.step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.cols));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.rows));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_offset));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_offset));
     args.push_back( make_pair( sizeof(cl_float), (void *)&scale));
 
-    size_t globalThreads[3] = {map_x.cols, map_x.rows, 1};
-    size_t localThreads[3]  = {32, 8, 1};
-    openCLExecuteKernel(clCxt, &build_warps, kernelName, globalThreads, localThreads, args, -1, -1);
+    size_t globalThreads[3] = { xmap.cols, xmap.rows, 1 };
+    size_t localThreads[3]  = { 32, 8, 1 };
+    openCLExecuteKernel(Context::getContext(), &build_warps, "buildWarpSphericalMaps", globalThreads, localThreads, args, -1, -1);
 }
 
+//////////////////////////////////////////////////////////////////////////////
+// buildWarpAffineMaps
 
 void cv::ocl::buildWarpAffineMaps(const Mat &M, bool inverse, Size dsize, oclMat &xmap, oclMat &ymap)
 {
-
     CV_Assert(M.rows == 2 && M.cols == 3);
+    CV_Assert(dsize.area());
 
     xmap.create(dsize, CV_32FC1);
     ymap.create(dsize, CV_32FC1);
@@ -194,29 +205,34 @@ void cv::ocl::buildWarpAffineMaps(const Mat &M, bool inverse, Size dsize, oclMat
         iM.convertTo(coeffsMat, coeffsMat.type());
     }
 
+    int xmap_step = xmap.step / xmap.elemSize(), xmap_offset = xmap.offset / xmap.elemSize();
+    int ymap_step = ymap.step / ymap.elemSize(), ymap_offset = ymap.offset / ymap.elemSize();
+
     oclMat coeffsOclMat(coeffsMat.reshape(1, 1));
 
-    Context *clCxt = Context::getContext();
-    string kernelName = "buildWarpAffineMaps";
     vector< pair<size_t, const void *> > args;
-
     args.push_back( make_pair( sizeof(cl_mem), (void *)&xmap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&ymap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&coeffsOclMat.data));
     args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.cols));
     args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.rows));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.step));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap.step));
-
-    size_t globalThreads[3] = {xmap.cols, xmap.rows, 1};
-    size_t localThreads[3]  = {32, 8, 1};
-    openCLExecuteKernel(clCxt, &build_warps, kernelName, globalThreads, localThreads, args, -1, -1);
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_offset));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_offset));
+
+    size_t globalThreads[3] = { xmap.cols, xmap.rows, 1 };
+    size_t localThreads[3]  = { 32, 8, 1 };
+    openCLExecuteKernel(Context::getContext(), &build_warps, "buildWarpAffineMaps", globalThreads, localThreads, args, -1, -1);
 }
 
+//////////////////////////////////////////////////////////////////////////////
+// buildWarpPerspectiveMaps
+
 void cv::ocl::buildWarpPerspectiveMaps(const Mat &M, bool inverse, Size dsize, oclMat &xmap, oclMat &ymap)
 {
-
     CV_Assert(M.rows == 3 && M.cols == 3);
+    CV_Assert(dsize.area() > 0);
 
     xmap.create(dsize, CV_32FC1);
     ymap.create(dsize, CV_32FC1);
@@ -235,19 +251,21 @@ void cv::ocl::buildWarpPerspectiveMaps(const Mat &M, bool inverse, Size dsize, o
 
     oclMat coeffsOclMat(coeffsMat.reshape(1, 1));
 
-    Context *clCxt = Context::getContext();
-    string kernelName = "buildWarpPerspectiveMaps";
-    vector< pair<size_t, const void *> > args;
+    int xmap_step = xmap.step / xmap.elemSize(), xmap_offset = xmap.offset / xmap.elemSize();
+    int ymap_step = ymap.step / ymap.elemSize(), ymap_offset = ymap.offset / ymap.elemSize();
 
+    vector< pair<size_t, const void *> > args;
     args.push_back( make_pair( sizeof(cl_mem), (void *)&xmap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&ymap.data));
     args.push_back( make_pair( sizeof(cl_mem), (void *)&coeffsOclMat.data));
     args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.cols));
     args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.rows));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap.step));
-    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap.step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_step));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&xmap_offset));
+    args.push_back( make_pair( sizeof(cl_int), (void *)&ymap_offset));
+
+    size_t globalThreads[3] = { xmap.cols, xmap.rows, 1 };
 
-    size_t globalThreads[3] = {xmap.cols, xmap.rows, 1};
-    size_t localThreads[3]  = {32, 8, 1};
-    openCLExecuteKernel(clCxt, &build_warps, kernelName, globalThreads, localThreads, args, -1, -1);
+    openCLExecuteKernel(Context::getContext(), &build_warps, "buildWarpPerspectiveMaps", globalThreads, NULL, args, -1, -1);
 }

diff --git a/modules/ocl/src/color.cpp b/modules/ocl/src/color.cpp
index 27c2fd5..6e2f403 100644 (file)
--- a/modules/ocl/src/color.cpp
+++ b/modules/ocl/src/color.cpp
@@ -50,254 +50,430 @@
 using namespace cv;
 using namespace cv::ocl;
 
-#ifndef CV_DESCALE
-#define CV_DESCALE(x, n) (((x) + (1 << ((n)-1))) >> (n))
-#endif
-
-#ifndef FLT_EPSILON
-#define FLT_EPSILON     1.192092896e-07F
-#endif
-
-namespace
-{
-
-void RGB2Gray_caller(const oclMat &src, oclMat &dst, int bidx)
+static void fromRGB_caller(const oclMat &src, oclMat &dst, int bidx, const std::string & kernelName,
+                           const std::string & additionalOptions = std::string(),
+                           const oclMat & data1 = oclMat(), const oclMat & data2 = oclMat())
 {
-    int channels = src.oclchannels();
     int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
     int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
 
     std::string build_options = format("-D DEPTH_%d", src.depth());
+    if (!additionalOptions.empty())
+        build_options += additionalOptions;
 
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&channels));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&bidx));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
 
-    size_t gt[3] = {src.cols, src.rows, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "RGB2Gray", gt, lt, args, -1, -1, build_options.c_str());
-}
+    if (!data1.empty())
+        args.push_back( make_pair( sizeof(cl_mem) , (void *)&data1.data ));
+    if (!data2.empty())
+        args.push_back( make_pair( sizeof(cl_mem) , (void *)&data2.data ));
 
-void Gray2RGB_caller(const oclMat &src, oclMat &dst)
-{
-    std::string build_options = format("-D DEPTH_%d", src.depth());
-    int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
-    int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
-
-    vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
-    args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
-    args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
-
-    size_t gt[3] = {src.cols, src.rows, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "Gray2RGB", gt, lt, args, -1, -1, build_options.c_str());
+    size_t gt[3] = { dst.cols, dst.rows, 1 }, lt[3] = { 16, 16, 1 };
+    openCLExecuteKernel(src.clCxt, &cvt_color, kernelName.c_str(), gt, lt, args, -1, -1, build_options.c_str());
 }
 
-void RGB2YUV_caller(const oclMat &src, oclMat &dst, int bidx)
+static void toRGB_caller(const oclMat &src, oclMat &dst, int bidx, const std::string & kernelName,
+                         const std::string & additionalOptions = std::string(), const oclMat & data = oclMat())
 {
-    int channels = src.oclchannels();
-    std::string build_options = format("-D DEPTH_%d", src.depth());
+    std::string build_options = format("-D DEPTH_%d -D dcn=%d", src.depth(), dst.channels());
+    if (!additionalOptions.empty())
+        build_options += additionalOptions;
+
     int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
     int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
 
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&channels));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&bidx));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
 
-    size_t gt[3] = {src.cols, src.rows, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "RGB2YUV", gt, lt, args, -1, -1, build_options.c_str());
+    if (!data.empty())
+        args.push_back( make_pair( sizeof(cl_mem) , (void *)&data.data ));
+
+    size_t gt[3] = { dst.cols, dst.rows, 1 }, lt[3] = { 16, 16, 1 };
+    openCLExecuteKernel(src.clCxt, &cvt_color, kernelName.c_str(), gt, lt, args, -1, -1, build_options.c_str());
 }
 
-void YUV2RGB_caller(const oclMat &src, oclMat &dst, int bidx)
+static void RGB_caller(const oclMat &src, oclMat &dst, bool reverse)
 {
-    int channels = src.oclchannels();
+    std::string build_options = format("-D DEPTH_%d -D dcn=%d -D scn=%d -D %s", src.depth(),
+                                       dst.channels(), src.channels(), reverse ? "REVERSE" : "ORDER");
     int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
     int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
 
-    std::string buildOptions = format("-D DEPTH_%d", src.depth());
-
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&channels));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&bidx));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
 
-    size_t gt[3] = {src.cols, src.rows, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "YUV2RGB", gt, lt, args, -1, -1, buildOptions.c_str());
+    size_t gt[3] = { dst.cols, dst.rows, 1 }, lt[3] = { 16, 16, 1 };
+    openCLExecuteKernel(src.clCxt, &cvt_color, "RGB", gt, lt, args, -1, -1, build_options.c_str());
 }
 
-void YUV2RGB_NV12_caller(const oclMat &src, oclMat &dst, int bidx)
+static void fromRGB5x5_caller(const oclMat &src, oclMat &dst, int bidx, int greenbits, const std::string & kernelName)
 {
-    std::string build_options = format("-D DEPTH_%d", src.depth());
-    int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
-    int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
+    std::string build_options = format("-D DEPTH_%d -D greenbits=%d -D dcn=%d",
+                                       src.depth(), greenbits, dst.channels());
+    int src_offset = src.offset >> 1, src_step = src.step >> 1;
+    int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step / dst.elemSize1();
 
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&bidx));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
 
-    size_t gt[3] = {dst.cols / 2, dst.rows / 2, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "YUV2RGBA_NV12", gt, lt, args, -1, -1, build_options.c_str());
+    size_t gt[3] = { dst.cols, dst.rows, 1 }, lt[3] = { 16, 16, 1 };
+    openCLExecuteKernel(src.clCxt, &cvt_color, kernelName.c_str(), gt, lt, args, -1, -1, build_options.c_str());
 }
 
-void RGB2YCrCb_caller(const oclMat &src, oclMat &dst, int bidx)
+static void toRGB5x5_caller(const oclMat &src, oclMat &dst, int bidx, int greenbits, const std::string & kernelName)
 {
-    int channels = src.oclchannels();
-    std::string build_options = format("-D DEPTH_%d", src.depth());
-    int src_offset = src.offset / src.elemSize1(), src_step = src.step1();
-    int dst_offset = dst.offset / dst.elemSize1(), dst_step = dst.step1();
+    std::string build_options = format("-D DEPTH_%d -D greenbits=%d -D scn=%d",
+                                       src.depth(), greenbits, src.channels());
+    int src_offset = (int)src.offset, src_step = (int)src.step;
+    int dst_offset = dst.offset >> 1, dst_step = dst.step >> 1;
 
     vector<pair<size_t , const void *> > args;
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&src.rows));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.cols));
+    args.push_back( make_pair( sizeof(cl_int) , (void *)&dst.rows));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_step));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_step));
-    args.push_back( make_pair( sizeof(cl_int) , (void *)&channels));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&bidx));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
     args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst.data));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&src_offset ));
     args.push_back( make_pair( sizeof(cl_int) , (void *)&dst_offset ));
 
-    size_t gt[3] = {src.cols, src.rows, 1}, lt[3] = {16, 16, 1};
-    openCLExecuteKernel(src.clCxt, &cvt_color, "RGB2YCrCb", gt, lt, args, -1, -1, build_options.c_str());
+    size_t gt[3] = { dst.cols, dst.rows, 1 }, lt[3] = { 16, 16, 1 };
+    openCLExecuteKernel(src.clCxt, &cvt_color, kernelName.c_str(), gt, lt, args, -1, -1, build_options.c_str());
 }
 
-void cvtColor_caller(const oclMat &src, oclMat &dst, int code, int dcn)
+static void cvtColor_caller(const oclMat &src, oclMat &dst, int code, int dcn)
 {
     Size sz = src.size();
-    int scn = src.oclchannels(), depth = src.depth(), bidx;
+    int scn = src.channels(), depth = src.depth(), bidx;
 
     CV_Assert(depth == CV_8U || depth == CV_16U || depth == CV_32F);
 
     switch (code)
     {
-        /*
-        case CV_BGR2BGRA: case CV_RGB2BGRA: case CV_BGRA2BGR:
-        case CV_RGBA2BGR: case CV_RGB2BGR: case CV_BGRA2RGBA:
-        case CV_BGR2BGR565: case CV_BGR2BGR555: case CV_RGB2BGR565: case CV_RGB2BGR555:
-        case CV_BGRA2BGR565: case CV_BGRA2BGR555: case CV_RGBA2BGR565: case CV_RGBA2BGR555:
-        case CV_BGR5652BGR: case CV_BGR5552BGR: case CV_BGR5652RGB: case CV_BGR5552RGB:
-        case CV_BGR5652BGRA: case CV_BGR5552BGRA: case CV_BGR5652RGBA: case CV_BGR5552RGBA:
-        */
-    case CV_BGR2GRAY:
-    case CV_BGRA2GRAY:
-    case CV_RGB2GRAY:
-    case CV_RGBA2GRAY:
+    case CV_BGR2BGRA: case CV_RGB2BGRA: case CV_BGRA2BGR:
+    case CV_RGBA2BGR: case CV_RGB2BGR: case CV_BGRA2RGBA:
+    {
+        CV_Assert(scn == 3 || scn == 4);
+        dcn = code == CV_BGR2BGRA || code == CV_RGB2BGRA || code == CV_BGRA2RGBA ? 4 : 3;
+        bool reverse = !(code == CV_BGR2BGRA || code == CV_BGRA2BGR);
+        dst.create(sz, CV_MAKE_TYPE(depth, dcn));
+        RGB_caller(src, dst, reverse);
+        break;
+    }
+    case CV_BGR2BGR565: case CV_BGR2BGR555: case CV_RGB2BGR565: case CV_RGB2BGR555:
+    case CV_BGRA2BGR565: case CV_BGRA2BGR555: case CV_RGBA2BGR565: case CV_RGBA2BGR555:
+    {
+        CV_Assert((scn == 3 || scn == 4) && depth == CV_8U );
+        bidx = code == CV_BGR2BGR565 || code == CV_BGR2BGR555 ||
+            code == CV_BGRA2BGR565 || code == CV_BGRA2BGR555 ? 0 : 2;
+        int greenbits = code == CV_BGR2BGR565 || code == CV_RGB2BGR565 ||
+            code == CV_BGRA2BGR565 || code == CV_RGBA2BGR565 ? 6 : 5;
+        dst.create(sz, CV_8UC2);
+        toRGB5x5_caller(src, dst, bidx, greenbits, "RGB2RGB5x5");
+        break;
+    }
+    case CV_BGR5652BGR: case CV_BGR5552BGR: case CV_BGR5652RGB: case CV_BGR5552RGB:
+    case CV_BGR5652BGRA: case CV_BGR5552BGRA: case CV_BGR5652RGBA: case CV_BGR5552RGBA:
+    {
+        dcn = code == CV_BGR5652BGRA || code == CV_BGR5552BGRA || code == CV_BGR5652RGBA || code == CV_BGR5552RGBA ? 4 : 3;
+        CV_Assert((dcn == 3 || dcn == 4) && scn == 2 && depth == CV_8U);
+        bidx = code == CV_BGR5652BGR || code == CV_BGR5552BGR ||
+            code == CV_BGR5652BGRA || code == CV_BGR5552BGRA ? 0 : 2;
+        int greenbits = code == CV_BGR5652BGR || code == CV_BGR5652RGB ||
+            code == CV_BGR5652BGRA || code == CV_BGR5652RGBA ? 6 : 5;
+        dst.create(sz, CV_MAKETYPE(depth, dcn));
+        fromRGB5x5_caller(src, dst, bidx, greenbits, "RGB5x52RGB");
+        break;
+    }
+    case CV_BGR5652GRAY: case CV_BGR5552GRAY:
+    {
+        CV_Assert(scn == 2 && depth == CV_8U);
+        dst.create(sz, CV_8UC1);
+        int greenbits = code == CV_BGR5652GRAY ? 6 : 5;
+        fromRGB5x5_caller(src, dst, -1, greenbits, "BGR5x52Gray");
+        break;
+    }
+    case CV_GRAY2BGR565: case CV_GRAY2BGR555:
+    {
+        CV_Assert(scn == 1 && depth == CV_8U);
+        dst.create(sz, CV_8UC2);
+        int greenbits = code == CV_GRAY2BGR565 ? 6 : 5;
+        toRGB5x5_caller(src, dst, -1, greenbits, "Gray2BGR5x5");
+        break;
+    }
+    case CV_RGB2GRAY: case CV_BGR2GRAY: case CV_RGBA2GRAY: case CV_BGRA2GRAY:
     {
         CV_Assert(scn == 3 || scn == 4);
         bidx = code == CV_BGR2GRAY || code == CV_BGRA2GRAY ? 0 : 2;
         dst.create(sz, CV_MAKETYPE(depth, 1));
-        RGB2Gray_caller(src, dst, bidx);
+        fromRGB_caller(src, dst, bidx, "RGB2Gray");
         break;
     }
-    case CV_GRAY2BGR:
-    case CV_GRAY2BGRA:
+    case CV_GRAY2BGR: case CV_GRAY2BGRA:
     {
         CV_Assert(scn == 1);
         dcn  = code == CV_GRAY2BGRA ? 4 : 3;
         dst.create(sz, CV_MAKETYPE(depth, dcn));
-        Gray2RGB_caller(src, dst);
+        toRGB_caller(src, dst, 0, "Gray2RGB");
         break;
     }
-    case CV_BGR2YUV:
-    case CV_RGB2YUV:
+    case CV_BGR2YUV: case CV_RGB2YUV:
     {
         CV_Assert(scn == 3 || scn == 4);
-        bidx = code == CV_RGB2YUV ? 0 : 2;
+        bidx = code == CV_BGR2YUV ? 0 : 2;
         dst.create(sz, CV_MAKETYPE(depth, 3));
-        RGB2YUV_caller(src, dst, bidx);
+        fromRGB_caller(src, dst, bidx, "RGB2YUV");
         break;
     }
-    case CV_YUV2BGR:
-    case CV_YUV2RGB:
+    case CV_YUV2BGR: case CV_YUV2RGB:
     {
-        CV_Assert(scn == 3 || scn == 4);
-        bidx = code == CV_YUV2RGB ? 0 : 2;
-        dst.create(sz, CV_MAKETYPE(depth, 3));
-        YUV2RGB_caller(src, dst, bidx);
+        if( dcn <= 0 )
+            dcn = 3;
+        CV_Assert(scn == 3 && (dcn == 3 || dcn == 4));
+        bidx = code == CV_YUV2BGR ? 0 : 2;
+        dst.create(sz, CV_MAKETYPE(depth, dcn));
+        toRGB_caller(src, dst, bidx, "YUV2RGB");
         break;
     }
-    case CV_YUV2RGB_NV12:
-    case CV_YUV2BGR_NV12:
-    case CV_YUV2RGBA_NV12:
-    case CV_YUV2BGRA_NV12:
+    case CV_YUV2RGB_NV12: case CV_YUV2BGR_NV12:
+    case CV_YUV2RGBA_NV12: case CV_YUV2BGRA_NV12:
     {
         CV_Assert(scn == 1);
         CV_Assert( sz.width % 2 == 0 && sz.height % 3 == 0 && depth == CV_8U );
-        dcn  = code == CV_YUV2BGRA_NV12 || code == CV_YUV2RGBA_NV12 ? 4 : 3;
+        dcn = code == CV_YUV2BGRA_NV12 || code == CV_YUV2RGBA_NV12 ? 4 : 3;
         bidx = code == CV_YUV2BGRA_NV12 || code == CV_YUV2BGR_NV12 ? 0 : 2;
 
         Size dstSz(sz.width, sz.height * 2 / 3);
         dst.create(dstSz, CV_MAKETYPE(depth, dcn));
-        YUV2RGB_NV12_caller(src, dst, bidx);
+        toRGB_caller(src, dst, bidx, "YUV2RGBA_NV12");
         break;
     }
-    case CV_BGR2YCrCb:
-    case CV_RGB2YCrCb:
+    case CV_BGR2YCrCb: case CV_RGB2YCrCb:
     {
         CV_Assert(scn == 3 || scn == 4);
         bidx = code == CV_BGR2YCrCb ? 0 : 2;
         dst.create(sz, CV_MAKETYPE(depth, 3));
-        RGB2YCrCb_caller(src, dst, bidx);
+        fromRGB_caller(src, dst, bidx, "RGB2YCrCb");
         break;
     }
-    case CV_YCrCb2BGR:
-    case CV_YCrCb2RGB:
+    case CV_YCrCb2BGR: case CV_YCrCb2RGB:
     {
+        if( dcn <= 0 )
+            dcn = 3;
+        CV_Assert(scn == 3 && (dcn == 3 || dcn == 4));
+        bidx = code == CV_YCrCb2BGR ? 0 : 2;
+        dst.create(sz, CV_MAKETYPE(depth, dcn));
+        toRGB_caller(src, dst, bidx, "YCrCb2RGB");
         break;
     }
-    /*
-    case CV_BGR5652GRAY: case CV_BGR5552GRAY:
-    case CV_GRAY2BGR565: case CV_GRAY2BGR555:
-    case CV_BGR2YCrCb: case CV_RGB2YCrCb:
     case CV_BGR2XYZ: case CV_RGB2XYZ:
+    {
+        CV_Assert(scn == 3 || scn == 4);
+        bidx = code == CV_BGR2XYZ ? 0 : 2;
+        dst.create(sz, CV_MAKE_TYPE(depth, 3));
+
+        Mat c;
+        if (depth == CV_32F)
+        {
+            float coeffs[] =
+            {
+                0.412453f, 0.357580f, 0.180423f,
+                0.212671f, 0.715160f, 0.072169f,
+                0.019334f, 0.119193f, 0.950227f
+            };
+            if (bidx == 0)
+            {
+                std::swap(coeffs[0], coeffs[2]);
+                std::swap(coeffs[3], coeffs[5]);
+                std::swap(coeffs[6], coeffs[8]);
+            }
+            Mat(1, 9, CV_32FC1, &coeffs[0]).copyTo(c);
+        }
+        else
+        {
+            int coeffs[] =
+            {
+                1689,    1465,    739,
+                871,     2929,    296,
+                79,      488,     3892
+            };
+            if (bidx == 0)
+            {
+                std::swap(coeffs[0], coeffs[2]);
+                std::swap(coeffs[3], coeffs[5]);
+                std::swap(coeffs[6], coeffs[8]);
+            }
+            Mat(1, 9, CV_32SC1, &coeffs[0]).copyTo(c);
+        }
+        oclMat oclCoeffs(c);
+
+        fromRGB_caller(src, dst, bidx, "RGB2XYZ", "", oclCoeffs);
+        break;
+    }
     case CV_XYZ2BGR: case CV_XYZ2RGB:
+    {
+        if (dcn <= 0)
+            dcn = 3;
+        CV_Assert(scn == 3 && (dcn == 3 || dcn == 4));
+        bidx = code == CV_XYZ2BGR ? 0 : 2;
+        dst.create(sz, CV_MAKE_TYPE(depth, dcn));
+
+        Mat c;
+        if (depth == CV_32F)
+        {
+            float coeffs[] =
+            {
+                3.240479f, -1.53715f, -0.498535f,
+                -0.969256f, 1.875991f, 0.041556f,
+                0.055648f, -0.204043f, 1.057311f
+            };
+            if (bidx == 0)
+            {
+                std::swap(coeffs[0], coeffs[6]);
+                std::swap(coeffs[1], coeffs[7]);
+                std::swap(coeffs[2], coeffs[8]);
+            }
+            Mat(1, 9, CV_32FC1, &coeffs[0]).copyTo(c);
+        }
+        else
+        {
+            int coeffs[] =
+            {
+                13273,  -6296,  -2042,
+                -3970,   7684,    170,
+                  228,   -836,   4331
+            };
+            if (bidx == 0)
+            {
+                std::swap(coeffs[0], coeffs[6]);
+                std::swap(coeffs[1], coeffs[7]);
+                std::swap(coeffs[2], coeffs[8]);
+            }
+            Mat(1, 9, CV_32SC1, &coeffs[0]).copyTo(c);
+        }
+        oclMat oclCoeffs(c);
+
+        toRGB_caller(src, dst, bidx, "XYZ2RGB", "", oclCoeffs);
+        break;
+    }
     case CV_BGR2HSV: case CV_RGB2HSV: case CV_BGR2HSV_FULL: case CV_RGB2HSV_FULL:
     case CV_BGR2HLS: case CV_RGB2HLS: case CV_BGR2HLS_FULL: case CV_RGB2HLS_FULL:
+    {
+        CV_Assert((scn == 3 || scn == 4) && (depth == CV_8U || depth == CV_32F));
+        bidx = code == CV_BGR2HSV || code == CV_BGR2HLS ||
+            code == CV_BGR2HSV_FULL || code == CV_BGR2HLS_FULL ? 0 : 2;
+        int hrange = depth == CV_32F ? 360 : code == CV_BGR2HSV || code == CV_RGB2HSV ||
+            code == CV_BGR2HLS || code == CV_RGB2HLS ? 180 : 256;
+        bool is_hsv = code == CV_BGR2HSV || code == CV_RGB2HSV || code == CV_BGR2HSV_FULL || code == CV_RGB2HSV_FULL;
+        dst.create(sz, CV_MAKETYPE(depth, 3));
+        std::string kernelName = std::string("RGB2") + (is_hsv ? "HSV" : "HLS");
+
+        if (is_hsv && depth == CV_8U)
+        {
+            static oclMat sdiv_data;
+            static oclMat hdiv_data180;
+            static oclMat hdiv_data256;
+            static int sdiv_table[256];
+            static int hdiv_table180[256];
+            static int hdiv_table256[256];
+            static volatile bool initialized180 = false, initialized256 = false;
+            volatile bool & initialized = hrange == 180 ? initialized180 : initialized256;
+
+            if (!initialized)
+            {
+                int * const hdiv_table = hrange == 180 ? hdiv_table180 : hdiv_table256, hsv_shift = 12;
+                oclMat & hdiv_data = hrange == 180 ? hdiv_data180 : hdiv_data256;
+
+                sdiv_table[0] = hdiv_table180[0] = hdiv_table256[0] = 0;
+
+                int v = 255 << hsv_shift;
+                if (!initialized180 && !initialized256)
+                {
+                    for(int i = 1; i < 256; i++ )
+                        sdiv_table[i] = saturate_cast<int>(v/(1.*i));
+                    sdiv_data.upload(Mat(1, 256, CV_32SC1, sdiv_table));
+                }
+
+                v = hrange << hsv_shift;
+                for (int i = 1; i < 256; i++ )
+                    hdiv_table[i] = saturate_cast<int>(v/(6.*i));
+
+                hdiv_data.upload(Mat(1, 256, CV_32SC1, hdiv_table));
+                initialized = true;
+            }
+
+            fromRGB_caller(src, dst, bidx, kernelName, format(" -D hrange=%d", hrange), sdiv_data, hrange == 256 ? hdiv_data256 : hdiv_data180);
+            return;
+        }
+
+        fromRGB_caller(src, dst, bidx, kernelName, format(" -D hscale=%f", hrange*(1.f/360.f)));
+        break;
+    }
     case CV_HSV2BGR: case CV_HSV2RGB: case CV_HSV2BGR_FULL: case CV_HSV2RGB_FULL:
     case CV_HLS2BGR: case CV_HLS2RGB: case CV_HLS2BGR_FULL: case CV_HLS2RGB_FULL:
-    */
+    {
+        if (dcn <= 0)
+            dcn = 3;
+        CV_Assert(scn == 3 && (dcn == 3 || dcn == 4) && (depth == CV_8U || depth == CV_32F));
+        bidx = code == CV_HSV2BGR || code == CV_HLS2BGR ||
+            code == CV_HSV2BGR_FULL || code == CV_HLS2BGR_FULL ? 0 : 2;
+        int hrange = depth == CV_32F ? 360 : code == CV_HSV2BGR || code == CV_HSV2RGB ||
+            code == CV_HLS2BGR || code == CV_HLS2RGB ? 180 : 255;
+        bool is_hsv = code == CV_HSV2BGR || code == CV_HSV2RGB ||
+                code == CV_HSV2BGR_FULL || code == CV_HSV2RGB_FULL;
+
+        dst.create(sz, CV_MAKETYPE(depth, dcn));
+
+        std::string kernelName = std::string(is_hsv ? "HSV" : "HLS") + "2RGB";
+        toRGB_caller(src, dst, bidx, kernelName, format(" -D hrange=%d -D hscale=%f", hrange, 6.f/hrange));
+        break;
+    }
+    case CV_RGBA2mRGBA: case CV_mRGBA2RGBA:
+        {
+            CV_Assert(scn == 4 && depth == CV_8U);
+            dst.create(sz, CV_MAKETYPE(depth, 4));
+            std::string kernelName = code == CV_RGBA2mRGBA ? "RGBA2mRGBA" : "mRGBA2RGBA";
+
+            fromRGB_caller(src, dst, 0, kernelName);
+            break;
+        }
     default:
         CV_Error( CV_StsBadFlag, "Unknown/unsupported color conversion code" );
     }
 }
-}
 
 void cv::ocl::cvtColor(const oclMat &src, oclMat &dst, int code, int dcn)
 {

diff --git a/modules/ocl/src/haar.cpp b/modules/ocl/src/haar.cpp
index b79ec0f..deff867 100644 (file)
--- a/modules/ocl/src/haar.cpp
+++ b/modules/ocl/src/haar.cpp
@@ -747,6 +747,15 @@ CvSeq *cv::ocl::OclCascadeClassifier::oclHaarDetectObjects( oclMat &gimg, CvMemS
         oclMat gsum(totalheight + 4, gimg.cols + 1, CV_32SC1);
         oclMat gsqsum(totalheight + 4, gimg.cols + 1, CV_32FC1);
 
+        int sdepth = 0;
+        if(Context::getContext()->supportsFeature(FEATURE_CL_DOUBLE))
+            sdepth = CV_64FC1;
+        else
+            sdepth = CV_32FC1;
+        sdepth = CV_MAT_DEPTH(sdepth);
+        int type = CV_MAKE_TYPE(sdepth, 1);
+        oclMat gsqsum_t(totalheight + 4, gimg.cols + 1, type);
+
         cl_mem stagebuffer;
         cl_mem nodebuffer;
         cl_mem candidatebuffer;
@@ -754,6 +763,7 @@ CvSeq *cv::ocl::OclCascadeClassifier::oclHaarDetectObjects( oclMat &gimg, CvMemS
         cv::Rect roi, roi2;
         cv::Mat imgroi, imgroisq;
         cv::ocl::oclMat resizeroi, gimgroi, gimgroisq;
+
         int grp_per_CU = 12;
 
         size_t blocksize = 8;
@@ -773,7 +783,7 @@ CvSeq *cv::ocl::OclCascadeClassifier::oclHaarDetectObjects( oclMat &gimg, CvMemS
             roi2 = Rect(0, 0, sz.width - 1, sz.height - 1);
             resizeroi = gimg1(roi2);
             gimgroi = gsum(roi);
-            gimgroisq = gsqsum(roi);
+            gimgroisq = gsqsum_t(roi);
             int width = gimgroi.cols - 1 - cascade->orig_window_size.width;
             int height = gimgroi.rows - 1 - cascade->orig_window_size.height;
             scaleinfo[i].width_height = (width << 16) | height;
@@ -787,8 +797,13 @@ CvSeq *cv::ocl::OclCascadeClassifier::oclHaarDetectObjects( oclMat &gimg, CvMemS
             scaleinfo[i].factor = factor;
             cv::ocl::resize(gimg, resizeroi, Size(sz.width - 1, sz.height - 1), 0, 0, INTER_LINEAR);
             cv::ocl::integral(resizeroi, gimgroi, gimgroisq);
+
             indexy += sz.height;
         }
+        if(gsqsum_t.depth() == CV_64F)
+            gsqsum_t.convertTo(gsqsum, CV_32FC1);
+        else
+            gsqsum = gsqsum_t;
 
         gcascade   = (GpuHidHaarClassifierCascade *)cascade->hid_cascade;
         stage      = (GpuHidHaarStageClassifier *)(gcascade + 1);
@@ -996,7 +1011,12 @@ CvSeq *cv::ocl::OclCascadeClassifier::oclHaarDetectObjects( oclMat &gimg, CvMemS
         int n_factors = 0;
         oclMat gsum;
         oclMat gsqsum;
-        cv::ocl::integral(gimg, gsum, gsqsum);
+        oclMat gsqsum_t;
+        cv::ocl::integral(gimg, gsum, gsqsum_t);
+        if(gsqsum_t.depth() == CV_64F)
+            gsqsum_t.convertTo(gsqsum, CV_32FC1);
+        else
+            gsqsum = gsqsum_t;
         CvSize sz;
         vector<CvSize> sizev;
         vector<float> scalev;
@@ -1271,12 +1291,16 @@ void cv::ocl::OclCascadeClassifierBuf::detectMultiScale(oclMat &gimg, CV_OUT std
             roi2 = Rect(0, 0, sz.width - 1, sz.height - 1);
             resizeroi = gimg1(roi2);
             gimgroi = gsum(roi);
-            gimgroisq = gsqsum(roi);
+            gimgroisq = gsqsum_t(roi);
 
             cv::ocl::resize(gimg, resizeroi, Size(sz.width - 1, sz.height - 1), 0, 0, INTER_LINEAR);
             cv::ocl::integral(resizeroi, gimgroi, gimgroisq);
             indexy += sz.height;
         }
+        if(gsqsum_t.depth() == CV_64F)
+            gsqsum_t.convertTo(gsqsum, CV_32FC1);
+        else
+            gsqsum = gsqsum_t;
 
         gcascade   = (GpuHidHaarClassifierCascade *)(cascade->hid_cascade);
         stage      = (GpuHidHaarStageClassifier *)(gcascade + 1);
@@ -1338,7 +1362,11 @@ void cv::ocl::OclCascadeClassifierBuf::detectMultiScale(oclMat &gimg, CV_OUT std
     }
     else
     {
-        cv::ocl::integral(gimg, gsum, gsqsum);
+        cv::ocl::integral(gimg, gsum, gsqsum_t);
+        if(gsqsum_t.depth() == CV_64F)
+            gsqsum_t.convertTo(gsqsum, CV_32FC1);
+        else
+            gsqsum = gsqsum_t;
 
         gcascade   = (GpuHidHaarClassifierCascade *)cascade->hid_cascade;
 
@@ -1564,6 +1592,7 @@ void cv::ocl::OclCascadeClassifierBuf::CreateFactorRelatedBufs(
             gimg1.release();
             gsum.release();
             gsqsum.release();
+            gsqsum_t.release();
         }
         else if (!(m_flags & CV_HAAR_SCALE_IMAGE) && (flags & CV_HAAR_SCALE_IMAGE))
         {
@@ -1638,6 +1667,16 @@ void cv::ocl::OclCascadeClassifierBuf::CreateFactorRelatedBufs(
         gsum.create(totalheight + 4, cols + 1, CV_32SC1);
         gsqsum.create(totalheight + 4, cols + 1, CV_32FC1);
 
+        int sdepth = 0;
+        if(Context::getContext()->supportsFeature(FEATURE_CL_DOUBLE))
+            sdepth = CV_64FC1;
+        else
+            sdepth = CV_32FC1;
+        sdepth = CV_MAT_DEPTH(sdepth);
+        int type = CV_MAKE_TYPE(sdepth, 1);
+
+        gsqsum_t.create(totalheight + 4, cols + 1, type);
+
         scaleinfo = (detect_piramid_info *)malloc(sizeof(detect_piramid_info) * loopcount);
         for( int i = 0; i < loopcount; i++ )
         {

diff --git a/modules/ocl/src/imgproc.cpp b/modules/ocl/src/imgproc.cpp
index 3539dfa..e8f42ed 100644 (file)
--- a/modules/ocl/src/imgproc.cpp
+++ b/modules/ocl/src/imgproc.cpp
@@ -195,9 +195,12 @@ namespace cv
                 return;
             }
 
-            CV_Assert(interpolation == INTER_LINEAR || interpolation == INTER_NEAREST
-                      || interpolation == INTER_CUBIC || interpolation == INTER_LANCZOS4);
-            CV_Assert((map1.type() == CV_16SC2 && !map2.data) || (map1.type() == CV_32FC2 && !map2.data) ||
+            if (map1.empty())
+                map1.swap(map2);
+
+            CV_Assert(interpolation == INTER_LINEAR || interpolation == INTER_NEAREST);
+            CV_Assert((map1.type() == CV_16SC2 && (map2.empty() || (map2.type() == CV_16UC1 || map2.type() == CV_16SC1)) ) ||
+                      (map1.type() == CV_32FC2 && !map2.data) ||
                       (map1.type() == CV_32FC1 && map2.type() == CV_32FC1));
             CV_Assert(!map2.data || map2.size() == map1.size());
             CV_Assert(borderType == BORDER_CONSTANT || borderType == BORDER_REPLICATE || borderType == BORDER_WRAP
@@ -212,18 +215,22 @@ namespace cv
                                    "BORDER_REFLECT_101", "BORDER_TRANSPARENT" };
 
             string kernelName = "remap";
-            if ( map1.type() == CV_32FC2 && !map2.data )
+            if (map1.type() == CV_32FC2 && map2.empty())
                 kernelName += "_32FC2";
-            else if (map1.type() == CV_16SC2 && !map2.data)
+            else if (map1.type() == CV_16SC2)
+            {
                 kernelName += "_16SC2";
+                if (!map2.empty())
+                    kernelName += "_16UC1";
+            }
             else if (map1.type() == CV_32FC1 && map2.type() == CV_32FC1)
                 kernelName += "_2_32FC1";
             else
                 CV_Error(CV_StsBadArg, "Unsupported map types");
 
             int ocn = dst.oclchannels();
-            size_t localThreads[3] = { 16, 16, 1};
-            size_t globalThreads[3] = { dst.cols, dst.rows, 1};
+            size_t localThreads[3] = { 256, 1, 1 };
+            size_t globalThreads[3] = { dst.cols, dst.rows, 1 };
 
             Mat scalar(1, 1, CV_MAKE_TYPE(dst.depth(), ocn), borderValue);
             std::string buildOptions = format("-D %s -D %s -D T=%s%s", interMap[interpolation],
@@ -232,9 +239,6 @@ namespace cv
             if (interpolation != INTER_NEAREST)
             {
                 int wdepth = std::max(CV_32F, dst.depth());
-                if (!supportsDouble)
-                    wdepth = std::min(CV_32F, wdepth);
-
                 buildOptions += format(" -D WT=%s%s -D convertToT=convert_%s%s%s -D convertToWT=convert_%s%s"
                                        " -D convertToWT2=convert_%s2 -D WT2=%s2",
                                        typeMap[wdepth], channelMap[ocn],
@@ -276,113 +280,213 @@ namespace cv
         ////////////////////////////////////////////////////////////////////////////////////////////
         // resize
 
-        static void resize_gpu( const oclMat &src, oclMat &dst, double fx, double fy, int interpolation)
+        static void computeResizeAreaTabs(int ssize, int dsize, double scale, int * const map_tab,
+                                          float * const alpha_tab, int * const ofs_tab)
         {
-            CV_Assert( (src.channels() == dst.channels()) );
-            Context *clCxt = src.clCxt;
-            float ifx = 1. / fx;
-            float ify = 1. / fy;
-            double ifx_d = 1. / fx;
-            double ify_d = 1. / fy;
-            int srcStep_in_pixel = src.step1() / src.oclchannels();
-            int srcoffset_in_pixel = src.offset / src.elemSize();
-            int dstStep_in_pixel = dst.step1() / dst.oclchannels();
-            int dstoffset_in_pixel = dst.offset / dst.elemSize();
-
-            string kernelName;
-            if (interpolation == INTER_LINEAR)
-                kernelName = "resizeLN";
-            else if (interpolation == INTER_NEAREST)
-                kernelName = "resizeNN";
-
-            //TODO: improve this kernel
+            int k = 0, dx = 0;
+            for ( ; dx < dsize; dx++)
+            {
+                ofs_tab[dx] = k;
+
+                double fsx1 = dx * scale;
+                double fsx2 = fsx1 + scale;
+                double cellWidth = std::min(scale, ssize - fsx1);
+
+                int sx1 = cvCeil(fsx1), sx2 = cvFloor(fsx2);
+
+                sx2 = std::min(sx2, ssize - 1);
+                sx1 = std::min(sx1, sx2);
+
+                if (sx1 - fsx1 > 1e-3)
+                {
+                    map_tab[k] = sx1 - 1;
+                    alpha_tab[k++] = (float)((sx1 - fsx1) / cellWidth);
+                }
+
+                for (int sx = sx1; sx < sx2; sx++)
+                {
+                    map_tab[k] = sx;
+                    alpha_tab[k++] = float(1.0 / cellWidth);
+                }
+
+                if (fsx2 - sx2 > 1e-3)
+                {
+                    map_tab[k] = sx2;
+                    alpha_tab[k++] = (float)(std::min(std::min(fsx2 - sx2, 1.), cellWidth) / cellWidth);
+                }
+            }
+            ofs_tab[dx] = k;
+        }
+
+        static void computeResizeAreaFastTabs(int * dmap_tab, int * smap_tab, int scale, int dcols, int scol)
+        {
+            for (int i = 0; i < dcols; ++i)
+                dmap_tab[i] = scale * i;
+
+            for (int i = 0, size = dcols * scale; i < size; ++i)
+                smap_tab[i] = std::min(scol - 1, i);
+        }
+
+        static void resize_gpu( const oclMat &src, oclMat &dst, double ifx, double ify, int interpolation)
+        {
+            float ifxf = (float)ifx, ifyf = (float)ify;
+            int src_step = src.step / src.elemSize(), src_offset = src.offset / src.elemSize();
+            int dst_step = dst.step / dst.elemSize(), dst_offset = dst.offset / dst.elemSize();
+            int ocn = dst.oclchannels(), depth = dst.depth();
+
+            const char * const interMap[] = { "NN", "LN", "CUBIC", "AREA", "LAN4" };
+            std::string kernelName = std::string("resize") + interMap[interpolation];
+
+            const char * const typeMap[] = { "uchar", "char", "ushort", "short", "int", "float", "double" };
+            const char * const channelMap[] = { "" , "", "2", "4", "4" };
+            std::string buildOption = format("-D %s -D T=%s%s", interMap[interpolation], typeMap[depth], channelMap[ocn]);
+
+            int wdepth = std::max(src.depth(), CV_32F);
+
+            // check if fx, fy is integer and then we have inter area fast mode
+            int iscale_x = saturate_cast<int>(ifx);
+            int iscale_y = saturate_cast<int>(ify);
+
+            bool is_area_fast = std::abs(ifx - iscale_x) < DBL_EPSILON &&
+                std::abs(ify - iscale_y) < DBL_EPSILON;
+            if (is_area_fast)
+                wdepth = std::max(src.depth(), CV_32S);
+
+            if (interpolation != INTER_NEAREST)
+            {
+                buildOption += format(" -D WT=%s -D WTV=%s%s -D convertToWTV=convert_%s%s -D convertToT=convert_%s%s%s",
+                                      typeMap[wdepth], typeMap[wdepth], channelMap[ocn],
+                                      typeMap[wdepth], channelMap[ocn],
+                                      typeMap[src.depth()], channelMap[ocn], src.depth() <= CV_32S ? "_sat_rte" : "");
+            }
+
             size_t blkSizeX = 16, blkSizeY = 16;
             size_t glbSizeX;
-            if (src.type() == CV_8UC1)
+            if (src.type() == CV_8UC1 && interpolation == INTER_LINEAR)
             {
                 size_t cols = (dst.cols + dst.offset % 4 + 3) / 4;
                 glbSizeX = cols % blkSizeX == 0 && cols != 0 ? cols : (cols / blkSizeX + 1) * blkSizeX;
             }
             else
-                glbSizeX = dst.cols % blkSizeX == 0 && dst.cols != 0 ? dst.cols : (dst.cols / blkSizeX + 1) * blkSizeX;
-
-            size_t glbSizeY = dst.rows % blkSizeY == 0 && dst.rows != 0 ? dst.rows : (dst.rows / blkSizeY + 1) * blkSizeY;
-            size_t globalThreads[3] = {glbSizeX, glbSizeY, 1};
-            size_t localThreads[3] = {blkSizeX, blkSizeY, 1};
+                glbSizeX = dst.cols;
 
-            vector< pair<size_t, const void *> > args;
-            if (interpolation == INTER_NEAREST)
+            oclMat alphaOcl, mapOcl, tabofsOcl;
+            if (interpolation == INTER_AREA)
             {
-                args.push_back( make_pair(sizeof(cl_mem), (void *)&dst.data));
-                args.push_back( make_pair(sizeof(cl_mem), (void *)&src.data));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dstoffset_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&srcoffset_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dstStep_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&srcStep_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&src.cols));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&src.rows));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dst.cols));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dst.rows));
-                if (src.clCxt->supportsFeature(FEATURE_CL_DOUBLE))
+                if (is_area_fast)
                 {
-                    args.push_back( make_pair(sizeof(cl_double), (void *)&ifx_d));
-                    args.push_back( make_pair(sizeof(cl_double), (void *)&ify_d));
+                    kernelName += "_FAST";
+                    int wdepth2 = std::max(CV_32F, src.depth());
+                    buildOption += format(" -D WT2V=%s%s -D convertToWT2V=convert_%s%s -D AREA_FAST -D XSCALE=%d -D YSCALE=%d -D SCALE=%f",
+                                          typeMap[wdepth2], channelMap[ocn], typeMap[wdepth2], channelMap[ocn],
+                                          iscale_x, iscale_y, 1.0f / (iscale_x * iscale_y));
+
+                    int smap_tab_size = dst.cols * iscale_x + dst.rows * iscale_y;
+                    AutoBuffer<int> dmap_tab(dst.cols + dst.rows), smap_tab(smap_tab_size);
+                    int * dxmap_tab = dmap_tab, * dymap_tab = dxmap_tab + dst.cols;
+                    int * sxmap_tab = smap_tab, * symap_tab = smap_tab + dst.cols * iscale_y;
+
+                    computeResizeAreaFastTabs(dxmap_tab, sxmap_tab, iscale_x, dst.cols, src.cols);
+                    computeResizeAreaFastTabs(dymap_tab, symap_tab, iscale_y, dst.rows, src.rows);
+
+                    tabofsOcl = oclMat(1, dst.cols + dst.rows, CV_32SC1, (void *)dmap_tab);
+                    mapOcl = oclMat(1, smap_tab_size, CV_32SC1, (void *)smap_tab);
                 }
                 else
                 {
-                    args.push_back( make_pair(sizeof(cl_float), (void *)&ifx));
-                    args.push_back( make_pair(sizeof(cl_float), (void *)&ify));
+                    Size ssize = src.size(), dsize = dst.size();
+                    int xytab_size = (ssize.width + ssize.height) << 1;
+                    int tabofs_size = dsize.height + dsize.width + 2;
+
+                    AutoBuffer<int> _xymap_tab(xytab_size), _xyofs_tab(tabofs_size);
+                    AutoBuffer<float> _xyalpha_tab(xytab_size);
+                    int * xmap_tab = _xymap_tab, * ymap_tab = _xymap_tab + (ssize.width << 1);
+                    float * xalpha_tab = _xyalpha_tab, * yalpha_tab = _xyalpha_tab + (ssize.width << 1);
+                    int * xofs_tab = _xyofs_tab, * yofs_tab = _xyofs_tab + dsize.width + 1;
+
+                    computeResizeAreaTabs(ssize.width, dsize.width, ifx, xmap_tab, xalpha_tab, xofs_tab);
+                    computeResizeAreaTabs(ssize.height, dsize.height, ify, ymap_tab, yalpha_tab, yofs_tab);
+
+                    // loading precomputed arrays to GPU
+                    alphaOcl = oclMat(1, xytab_size, CV_32FC1, (void *)_xyalpha_tab);
+                    mapOcl = oclMat(1, xytab_size, CV_32SC1, (void *)_xymap_tab);
+                    tabofsOcl = oclMat(1, tabofs_size, CV_32SC1, (void *)_xyofs_tab);
                 }
             }
+
+            size_t globalThreads[3] = { glbSizeX, dst.rows, 1 };
+            size_t localThreads[3] = { blkSizeX, blkSizeY, 1 };
+
+            std::vector< std::pair<size_t, const void *> > args;
+            args.push_back( make_pair(sizeof(cl_mem), (void *)&dst.data));
+            args.push_back( make_pair(sizeof(cl_mem), (void *)&src.data));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&dst_offset));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&src_offset));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&dst_step));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&src_step));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&src.cols));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&src.rows));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&dst.cols));
+            args.push_back( make_pair(sizeof(cl_int), (void *)&dst.rows));
+
+            if (wdepth == CV_64F)
+            {
+                args.push_back( make_pair(sizeof(cl_double), (void *)&ifx));
+                args.push_back( make_pair(sizeof(cl_double), (void *)&ify));
+            }
             else
             {
-                args.push_back( make_pair(sizeof(cl_mem), (void *)&dst.data));
-                args.push_back( make_pair(sizeof(cl_mem), (void *)&src.data));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dstoffset_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&srcoffset_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dstStep_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&srcStep_in_pixel));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&src.cols));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&src.rows));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dst.cols));
-                args.push_back( make_pair(sizeof(cl_int), (void *)&dst.rows));
-                args.push_back( make_pair(sizeof(cl_float), (void *)&ifx));
-                args.push_back( make_pair(sizeof(cl_float), (void *)&ify));
+                args.push_back( make_pair(sizeof(cl_float), (void *)&ifxf));
+                args.push_back( make_pair(sizeof(cl_float), (void *)&ifyf));
             }
 
-            openCLExecuteKernel(clCxt, &imgproc_resize, kernelName, globalThreads, localThreads, args, src.oclchannels(), src.depth());
+            // precomputed tabs
+            if (!tabofsOcl.empty())
+                args.push_back( make_pair(sizeof(cl_mem), (void *)&tabofsOcl.data));
+
+            if (!mapOcl.empty())
+                args.push_back( make_pair(sizeof(cl_mem), (void *)&mapOcl.data));
+
+            if (!alphaOcl.empty())
+                args.push_back( make_pair(sizeof(cl_mem), (void *)&alphaOcl.data));
+
+            ocn = interpolation == INTER_LINEAR ? ocn : -1;
+            depth = interpolation == INTER_LINEAR ? depth : -1;
+
+            openCLExecuteKernel(src.clCxt, &imgproc_resize, kernelName, globalThreads, localThreads, args,
+                                ocn, depth, buildOption.c_str());
         }
 
-        void resize(const oclMat &src, oclMat &dst, Size dsize,
-                    double fx, double fy, int interpolation)
+        void resize(const oclMat &src, oclMat &dst, Size dsize, double fx, double fy, int interpolation)
         {
+            if (!src.clCxt->supportsFeature(FEATURE_CL_DOUBLE) && src.depth() == CV_64F)
+            {
+                CV_Error(CV_OpenCLDoubleNotSupported, "Selected device does not support double");
+                return;
+            }
+
             CV_Assert(src.type() == CV_8UC1 || src.type() == CV_8UC3 || src.type() == CV_8UC4
                       || src.type() == CV_32FC1 || src.type() == CV_32FC3 || src.type() == CV_32FC4);
-            CV_Assert(interpolation == INTER_LINEAR || interpolation == INTER_NEAREST);
-            CV_Assert( src.size().area() > 0 );
-            CV_Assert( !(dsize == Size()) || (fx > 0 && fy > 0) );
-
-            if (!(dsize == Size()) && (fx > 0 && fy > 0))
-                if (dsize.width != (int)(src.cols * fx) || dsize.height != (int)(src.rows * fy))
-                    CV_Error(CV_StsUnmatchedSizes, "invalid dsize and fx, fy!");
+            CV_Assert(dsize.area() > 0 || (fx > 0 && fy > 0));
 
-            if ( dsize == Size() )
+            if (dsize.area() == 0)
+            {
                 dsize = Size(saturate_cast<int>(src.cols * fx), saturate_cast<int>(src.rows * fy));
+                CV_Assert(dsize.area() > 0);
+            }
             else
             {
                 fx = (double)dsize.width / src.cols;
                 fy = (double)dsize.height / src.rows;
             }
 
-            dst.create(dsize, src.type());
+            double inv_fy = 1 / fy, inv_fx = 1 / fx;
+            CV_Assert(interpolation == INTER_LINEAR || interpolation == INTER_NEAREST ||
+                      (interpolation == INTER_AREA && inv_fx >= 1 && inv_fy >= 1));
 
-            if ( interpolation == INTER_NEAREST || interpolation == INTER_LINEAR )
-            {
-                resize_gpu( src, dst, fx, fy, interpolation);
-                return;
-            }
+            dst.create(dsize, src.type());
 
-            CV_Error(CV_StsUnsupportedFormat, "Non-supported interpolation method");
+            resize_gpu( src, dst, inv_fx, inv_fy, interpolation);
         }
 
         ////////////////////////////////////////////////////////////////////////
@@ -777,7 +881,7 @@ namespace cv
         ////////////////////////////////////////////////////////////////////////
         // integral
 
-        void integral(const oclMat &src, oclMat &sum, oclMat &sqsum)
+        void integral(const oclMat &src, oclMat &sum, oclMat &sqsum, int sdepth)
         {
             CV_Assert(src.type() == CV_8UC1);
             if (!src.clCxt->supportsFeature(ocl::FEATURE_CL_DOUBLE) && src.depth() == CV_64F)
@@ -786,6 +890,11 @@ namespace cv
                 return;
             }
 
+            if( sdepth <= 0 )
+                sdepth = CV_32S;
+            sdepth = CV_MAT_DEPTH(sdepth);
+            int type = CV_MAKE_TYPE(sdepth, 1);
+
             int vlen = 4;
             int offset = src.offset / vlen;
             int pre_invalid = src.offset % vlen;
@@ -793,17 +902,26 @@ namespace cv
 
             oclMat t_sum , t_sqsum;
             int w = src.cols + 1, h = src.rows + 1;
-            int depth = src.depth() == CV_8U ? CV_32S : CV_64F;
-            int type = CV_MAKE_TYPE(depth, 1);
+
+            char build_option[250];
+            if(Context::getContext()->supportsFeature(ocl::FEATURE_CL_DOUBLE))
+            {
+                t_sqsum.create(src.cols, src.rows, CV_64FC1);
+                sqsum.create(h, w, CV_64FC1);
+                sprintf(build_option, "-D TYPE=double -D TYPE4=double4 -D convert_TYPE4=convert_double4");
+            }
+            else
+            {
+                t_sqsum.create(src.cols, src.rows, CV_32FC1);
+                sqsum.create(h, w, CV_32FC1);
+                sprintf(build_option, "-D TYPE=float -D TYPE4=float4 -D convert_TYPE4=convert_float4");
+            }
 
             t_sum.create(src.cols, src.rows, type);
             sum.create(h, w, type);
 
-            t_sqsum.create(src.cols, src.rows, CV_32FC1);
-            sqsum.create(h, w, CV_32FC1);
-
-            int sum_offset = sum.offset / vlen;
-            int sqsum_offset = sqsum.offset / vlen;
+            int sum_offset = sum.offset / sum.elemSize();
+            int sqsum_offset = sqsum.offset / sqsum.elemSize();
 
             vector<pair<size_t , const void *> > args;
             args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data ));
@@ -815,8 +933,9 @@ namespace cv
             args.push_back( make_pair( sizeof(cl_int) , (void *)&src.cols ));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&src.step ));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sum.step));
+            args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sqsum.step));
             size_t gt[3] = {((vcols + 1) / 2) * 256, 1, 1}, lt[3] = {256, 1, 1};
-            openCLExecuteKernel(src.clCxt, &imgproc_integral, "integral_cols", gt, lt, args, -1, depth);
+            openCLExecuteKernel(src.clCxt, &imgproc_integral, "integral_cols", gt, lt, args, -1, sdepth, build_option);
 
             args.clear();
             args.push_back( make_pair( sizeof(cl_mem) , (void *)&t_sum.data ));
@@ -826,15 +945,16 @@ namespace cv
             args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sum.rows ));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sum.cols ));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sum.step ));
+            args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sqsum.step));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sum.step));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sqsum.step));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sum_offset));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sqsum_offset));
             size_t gt2[3] = {t_sum.cols  * 32, 1, 1}, lt2[3] = {256, 1, 1};
-            openCLExecuteKernel(src.clCxt, &imgproc_integral, "integral_rows", gt2, lt2, args, -1, depth);
+            openCLExecuteKernel(src.clCxt, &imgproc_integral, "integral_rows", gt2, lt2, args, -1, sdepth, build_option);
         }
 
-        void integral(const oclMat &src, oclMat &sum)
+        void integral(const oclMat &src, oclMat &sum, int sdepth)
         {
             CV_Assert(src.type() == CV_8UC1);
             int vlen = 4;
@@ -842,10 +962,13 @@ namespace cv
             int pre_invalid = src.offset % vlen;
             int vcols = (pre_invalid + src.cols + vlen - 1) / vlen;
 
+            if( sdepth <= 0 )
+                sdepth = CV_32S;
+            sdepth = CV_MAT_DEPTH(sdepth);
+            int type = CV_MAKE_TYPE(sdepth, 1);
+
             oclMat t_sum;
             int w = src.cols + 1, h = src.rows + 1;
-            int depth = src.depth() == CV_8U ? CV_32S : CV_32F;
-            int type = CV_MAKE_TYPE(depth, 1);
 
             t_sum.create(src.cols, src.rows, type);
             sum.create(h, w, type);
@@ -861,7 +984,7 @@ namespace cv
             args.push_back( make_pair( sizeof(cl_int) , (void *)&src.step ));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&t_sum.step));
             size_t gt[3] = {((vcols + 1) / 2) * 256, 1, 1}, lt[3] = {256, 1, 1};
-            openCLExecuteKernel(src.clCxt, &imgproc_integral_sum, "integral_sum_cols", gt, lt, args, -1, depth);
+            openCLExecuteKernel(src.clCxt, &imgproc_integral_sum, "integral_sum_cols", gt, lt, args, -1, sdepth);
 
             args.clear();
             args.push_back( make_pair( sizeof(cl_mem) , (void *)&t_sum.data ));
@@ -872,7 +995,7 @@ namespace cv
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sum.step));
             args.push_back( make_pair( sizeof(cl_int) , (void *)&sum_offset));
             size_t gt2[3] = {t_sum.cols  * 32, 1, 1}, lt2[3] = {256, 1, 1};
-            openCLExecuteKernel(src.clCxt, &imgproc_integral_sum, "integral_sum_rows", gt2, lt2, args, -1, depth);
+            openCLExecuteKernel(src.clCxt, &imgproc_integral_sum, "integral_sum_rows", gt2, lt2, args, -1, sdepth);
         }
 
         /////////////////////// corner //////////////////////////////

diff --git a/modules/ocl/src/kmeans.cpp b/modules/ocl/src/kmeans.cpp
index 58a68a7..31fb250 100644 (file)
--- a/modules/ocl/src/kmeans.cpp
+++ b/modules/ocl/src/kmeans.cpp
@@ -160,63 +160,66 @@ static void generateCentersPP(const Mat& _data, Mat& _out_centers,
     }
 }
 
-void cv::ocl::distanceToCenters(oclMat &dists, oclMat &labels, const oclMat &src, const oclMat &centers, int distType, const oclMat &indices)
+void cv::ocl::distanceToCenters(const oclMat &src, const oclMat &centers, Mat &dists, Mat &labels, int distType)
 {
-    CV_Assert(src.cols*src.oclchannels() == centers.cols*centers.oclchannels());
+    CV_Assert(src.cols * src.channels() == centers.cols * centers.channels());
     CV_Assert(src.depth() == CV_32F && centers.depth() == CV_32F);
-    bool is_label_row_major = false;
-    ensureSizeIsEnough(1, src.rows, CV_32FC1, dists);
-    if(labels.empty() || (!labels.empty() && labels.rows == src.rows && labels.cols == 1))
-    {
-        ensureSizeIsEnough(src.rows, 1, CV_32SC1, labels);
-        is_label_row_major = true;
-    }
     CV_Assert(distType == NORM_L1 || distType == NORM_L2SQR);
 
-    std::stringstream build_opt_ss;
-    build_opt_ss
-        << (distType == NORM_L1 ? "-D L1_DIST" : "-D L2SQR_DIST")
-        << (indices.empty() ? "" : " -D USE_INDEX");
-
-    String build_opt = build_opt_ss.str();
+    dists.create(src.rows, 1, CV_32FC1);
+    labels.create(src.rows, 1, CV_32SC1);
 
-    const int src_step = (int)(src.oclchannels() * src.step / src.elemSize());
-    const int centers_step = (int)(centers.oclchannels() * centers.step / centers.elemSize());
-
-    const int colsNumb = centers.cols*centers.oclchannels();
-
-    const int label_step   = is_label_row_major ? (int)(labels.step / labels.elemSize()) : 1;
-    String kernelname = "distanceToCenters";
-
-    const int number_of_input = indices.empty() ? src.rows : indices.size().area();
+    std::stringstream build_opt_ss;
+    build_opt_ss << (distType == NORM_L1 ? "-D L1_DIST" : "-D L2SQR_DIST");
 
-    const int src_offset = (int)src.offset/src.elemSize();
-    const int centers_offset = (int)centers.offset/centers.elemSize();
+    int src_step = src.step / src.elemSize1();
+    int centers_step = centers.step / centers.elemSize1();
+    int feature_width = centers.cols * centers.oclchannels();
+    int src_offset = src.offset / src.elemSize1();
+    int centers_offset = centers.offset / centers.elemSize1();
 
-    size_t globalThreads[3] = {number_of_input, 1, 1};
+    int all_dist_count = src.rows * centers.rows;
+    oclMat all_dist(1, all_dist_count, CV_32FC1);
 
     vector<pair<size_t, const void *> > args;
     args.push_back(make_pair(sizeof(cl_mem), (void *)&src.data));
     args.push_back(make_pair(sizeof(cl_mem), (void *)&centers.data));
-    if(!indices.empty())
-    {
-        args.push_back(make_pair(sizeof(cl_mem), (void *)&indices.data));
-    }
-    args.push_back(make_pair(sizeof(cl_mem), (void *)&labels.data));
-    args.push_back(make_pair(sizeof(cl_mem), (void *)&dists.data));
-    args.push_back(make_pair(sizeof(cl_int), (void *)&colsNumb));
+    args.push_back(make_pair(sizeof(cl_mem), (void *)&all_dist.data));
+
+    args.push_back(make_pair(sizeof(cl_int), (void *)&feature_width));
     args.push_back(make_pair(sizeof(cl_int), (void *)&src_step));
     args.push_back(make_pair(sizeof(cl_int), (void *)&centers_step));
-    args.push_back(make_pair(sizeof(cl_int), (void *)&label_step));
-    args.push_back(make_pair(sizeof(cl_int), (void *)&number_of_input));
+    args.push_back(make_pair(sizeof(cl_int), (void *)&src.rows));
     args.push_back(make_pair(sizeof(cl_int), (void *)&centers.rows));
+
     args.push_back(make_pair(sizeof(cl_int), (void *)&src_offset));
     args.push_back(make_pair(sizeof(cl_int), (void *)&centers_offset));
 
+    size_t globalThreads[3] = { all_dist_count, 1, 1 };
+
     openCLExecuteKernel(Context::getContext(), &kmeans_kernel,
-        kernelname, globalThreads, NULL, args, -1, -1, build_opt.c_str());
+                        "distanceToCenters", globalThreads, NULL, args, -1, -1, build_opt_ss.str().c_str());
+
+    Mat all_dist_cpu;
+    all_dist.download(all_dist_cpu);
+
+    for (int i = 0; i < src.rows; ++i)
+    {
+        Point p;
+        double minVal;
+
+        Rect roi(i * centers.rows, 0, centers.rows, 1);
+        Mat hdr(all_dist_cpu, roi);
+
+        cv::minMaxLoc(hdr, &minVal, NULL, &p);
+
+        dists.at<float>(i, 0) = static_cast<float>(minVal);
+        labels.at<int>(i, 0) = p.x;
+    }
 }
+
 ///////////////////////////////////k - means /////////////////////////////////////////////////////////
+
 double cv::ocl::kmeans(const oclMat &_src, int K, oclMat &_bestLabels,
                        TermCriteria criteria, int attempts, int flags, oclMat &_centers)
 {
@@ -429,28 +432,19 @@ double cv::ocl::kmeans(const oclMat &_src, int K, oclMat &_bestLabels,
                 break;
 
             // assign labels
-            oclMat _dists(1, N, CV_64F);
-
-            _bestLabels.upload(_labels);
+            Mat dists(1, N, CV_64F);
             _centers.upload(centers);
+            distanceToCenters(_src, _centers, dists, _labels);
+            _bestLabels.upload(_labels);
 
-            distanceToCenters(_dists, _bestLabels, _src, _centers);
-
-            Mat dists;
-            _dists.download(dists);
-            _bestLabels.download(_labels);
             float* dist = dists.ptr<float>(0);
             compactness = 0;
             for( i = 0; i < N; i++ )
-            {
-                    compactness += (double)dist[i];
-            }
+                compactness += (double)dist[i];
         }
 
         if( compactness < best_compactness )
-        {
             best_compactness = compactness;
-        }
     }
 
     return best_compactness;

diff --git a/modules/ocl/src/match_template.cpp b/modules/ocl/src/match_template.cpp
index afd68ff..28397b6 100644 (file)
--- a/modules/ocl/src/match_template.cpp
+++ b/modules/ocl/src/match_template.cpp
@@ -245,12 +245,15 @@ namespace cv
         void matchTemplate_CCORR_NORMED(
             const oclMat &image, const oclMat &templ, oclMat &result, MatchTemplateBuf &buf)
         {
+            cv::ocl::oclMat temp;
             matchTemplate_CCORR(image, templ, result, buf);
             buf.image_sums.resize(1);
             buf.image_sqsums.resize(1);
-
-            integral(image.reshape(1), buf.image_sums[0], buf.image_sqsums[0]);
-
+            integral(image.reshape(1), buf.image_sums[0], temp);
+            if(temp.depth() == CV_64F)
+                temp.convertTo(buf.image_sqsums[0], CV_32FC1);
+            else
+                buf.image_sqsums[0] = temp;
             unsigned long long templ_sqsum = (unsigned long long)sqrSum(templ.reshape(1))[0];
 
             Context *clCxt = image.clCxt;
@@ -416,7 +419,12 @@ namespace cv
             {
                 buf.image_sums.resize(1);
                 buf.image_sqsums.resize(1);
-                integral(image, buf.image_sums[0], buf.image_sqsums[0]);
+                cv::ocl::oclMat temp;
+                integral(image, buf.image_sums[0], temp);
+                if(temp.depth() == CV_64F)
+                    temp.convertTo(buf.image_sqsums[0], CV_32FC1);
+                else
+                    buf.image_sqsums[0] = temp;
 
                 templ_sum[0]   = (float)sum(templ)[0];
 
@@ -452,10 +460,14 @@ namespace cv
                 templ_sum   *= scale;
                 buf.image_sums.resize(buf.images.size());
                 buf.image_sqsums.resize(buf.images.size());
-
+                cv::ocl::oclMat temp;
                 for(int i = 0; i < image.oclchannels(); i ++)
                 {
-                    integral(buf.images[i], buf.image_sums[i], buf.image_sqsums[i]);
+                    integral(buf.images[i], buf.image_sums[i], temp);
+                    if(temp.depth() == CV_64F)
+                        temp.convertTo(buf.image_sqsums[i], CV_32FC1);
+                    else
+                        buf.image_sqsums[i] = temp;
                 }
 
                 switch(image.oclchannels())

diff --git a/modules/ocl/src/opencl/arithm_LUT.cl b/modules/ocl/src/opencl/arithm_LUT.cl
index 658e1f4..30407bb 100644 (file)
--- a/modules/ocl/src/opencl/arithm_LUT.cl
+++ b/modules/ocl/src/opencl/arithm_LUT.cl
@@ -34,9 +34,13 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
 __kernel void LUT_C1( __global const srcT * src, __global const dstT *lut,
       __global dstT *dst,

diff --git a/modules/ocl/src/opencl/arithm_absdiff_nonsaturate.cl b/modules/ocl/src/opencl/arithm_absdiff_nonsaturate.cl
index fcf3874..e03fa69 100644 (file)
--- a/modules/ocl/src/opencl/arithm_absdiff_nonsaturate.cl
+++ b/modules/ocl/src/opencl/arithm_absdiff_nonsaturate.cl
@@ -44,11 +44,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 
@@ -65,12 +65,16 @@ __kernel void arithm_absdiff_nonsaturate_binary(__global srcT *src1, int src1_st
         int src1_index = mad24(y, src1_step, x + src1_offset);
         int src2_index = mad24(y, src2_step, x + src2_offset);
         int dst_index  = mad24(y, dst_step, x + dst_offset);
+#ifdef INTEL_DEVICE //workaround for intel compiler bug
+        if(src1_index >= 0 && src2_index >= 0)
+#endif
+        {
+            dstT t0 = convertToDstT(src1[src1_index]);
+            dstT t1 = convertToDstT(src2[src2_index]);
+            dstT t2 = t0 - t1;
 
-        dstT t0 = convertToDstT(src1[src1_index]);
-        dstT t1 = convertToDstT(src2[src2_index]);
-        dstT t2 = t0 - t1;
-
-        dst[dst_index] = t2 >= (dstT)(0) ? t2 : -t2;
+            dst[dst_index] = t2 >= (dstT)(0) ? t2 : -t2;
+        }
     }
 }
 
@@ -85,9 +89,13 @@ __kernel void arithm_absdiff_nonsaturate(__global srcT *src1, int src1_step, int
     {
         int src1_index = mad24(y, src1_step, x + src1_offset);
         int dst_index  = mad24(y, dst_step, x + dst_offset);
+#ifdef INTEL_DEVICE //workaround for intel compiler bug
+        if(src1_index >= 0)
+#endif
+        {
+            dstT t0 = convertToDstT(src1[src1_index]);
 
-        dstT t0 = convertToDstT(src1[src1_index]);
-
-        dst[dst_index] = t0 >= (dstT)(0) ? t0 : -t0;
+            dst[dst_index] = t0 >= (dstT)(0) ? t0 : -t0;
+        }
     }
 }

diff --git a/modules/ocl/src/opencl/arithm_add.cl b/modules/ocl/src/opencl/arithm_add.cl
index a73b65d..04262b8 100644 (file)
--- a/modules/ocl/src/opencl/arithm_add.cl
+++ b/modules/ocl/src/opencl/arithm_add.cl
@@ -44,11 +44,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_addWeighted.cl b/modules/ocl/src/opencl/arithm_addWeighted.cl
index 8272806..872ee85 100644 (file)
--- a/modules/ocl/src/opencl/arithm_addWeighted.cl
+++ b/modules/ocl/src/opencl/arithm_addWeighted.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_add_mask.cl b/modules/ocl/src/opencl/arithm_add_mask.cl
index ea96d8a..b115d9b 100644 (file)
--- a/modules/ocl/src/opencl/arithm_add_mask.cl
+++ b/modules/ocl/src/opencl/arithm_add_mask.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_add_scalar.cl b/modules/ocl/src/opencl/arithm_add_scalar.cl
index b82eff5..05ea48d 100644 (file)
--- a/modules/ocl/src/opencl/arithm_add_scalar.cl
+++ b/modules/ocl/src/opencl/arithm_add_scalar.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_add_scalar_mask.cl b/modules/ocl/src/opencl/arithm_add_scalar_mask.cl
index 0762b19..a8b9657 100644 (file)
--- a/modules/ocl/src/opencl/arithm_add_scalar_mask.cl
+++ b/modules/ocl/src/opencl/arithm_add_scalar_mask.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_bitwise_binary_scalar_mask.cl b/modules/ocl/src/opencl/arithm_bitwise_binary_scalar_mask.cl
index 03f46cc..756f201 100644 (file)
--- a/modules/ocl/src/opencl/arithm_bitwise_binary_scalar_mask.cl
+++ b/modules/ocl/src/opencl/arithm_bitwise_binary_scalar_mask.cl
@@ -43,14 +43,6 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
-#pragma OPENCL EXTENSION cl_amd_fp64:enable
-#endif
-#endif
-
 //////////////////////////////////////////////////////////////////////////////////////////////////////
 ////////////////////////////////////////////bitwise_binary////////////////////////////////////////////
 //////////////////////////////////////////////////////////////////////////////////////////////////////

diff --git a/modules/ocl/src/opencl/arithm_bitwise_not.cl b/modules/ocl/src/opencl/arithm_bitwise_not.cl
index 5bc1839..b6f76d6 100644 (file)
--- a/modules/ocl/src/opencl/arithm_bitwise_not.cl
+++ b/modules/ocl/src/opencl/arithm_bitwise_not.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_cartToPolar.cl b/modules/ocl/src/opencl/arithm_cartToPolar.cl
index e37818c..c65f899 100644 (file)
--- a/modules/ocl/src/opencl/arithm_cartToPolar.cl
+++ b/modules/ocl/src/opencl/arithm_cartToPolar.cl
@@ -43,98 +43,99 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-    #pragma OPENCL EXTENSION cl_khr_fp64:enable
-    #define CV_PI   3.1415926535897932384626433832795
-    #ifndef DBL_EPSILON
-        #define DBL_EPSILON 0x1.0p-52
-    #endif
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
+#define CV_PI M_PI
 #else
-    #define CV_PI   3.1415926535897932384626433832795f
-    #ifndef DBL_EPSILON
-        #define DBL_EPSILON 0x1.0p-52f
-    #endif
+#define CV_PI M_PI_F
 #endif
 
-
 __kernel void arithm_cartToPolar_D5 (__global float *src1, int src1_step, int src1_offset,
                                      __global float *src2, int src2_step, int src2_offset,
-                                     __global float *dst1, int dst1_step, int dst1_offset, //magnitude
-                                     __global float *dst2, int dst2_step, int dst2_offset, //cartToPolar
-                                     int rows, int cols, int angInDegree)
+                                     __global float *dst1, int dst1_step, int dst1_offset, // magnitude
+                                     __global float *dst2, int dst2_step, int dst2_offset, // cartToPolar
+                                     int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 2) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 2) + src2_offset);
+        int src1_index = mad24(y, src1_step, x + src1_offset);
+        int src2_index = mad24(y, src2_step, x + src2_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 2) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 2) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        float x = *((__global float *)((__global char *)src1 + src1_index));
-        float y = *((__global float *)((__global char *)src2 + src2_index));
+        float x = src1[src1_index];
+        float y = src2[src2_index];
 
         float x2 = x * x;
         float y2 = y * y;
 
         float magnitude = sqrt(x2 + y2);
-        float cartToPolar;
 
         float tmp = y >= 0 ? 0 : CV_PI*2;
         tmp = x < 0 ? CV_PI : tmp;
 
         float tmp1 = y >= 0 ? CV_PI*0.5f : CV_PI*1.5f;
-        cartToPolar = y2 <= x2 ? x*y/(x2 + 0.28f*y2 + DBL_EPSILON)  + tmp :
-                                 tmp1 - x*y/(y2 + 0.28f*x2 + DBL_EPSILON);
+        float cartToPolar = y2 <= x2 ? x*y/(x2 + 0.28f*y2 + FLT_EPSILON) + tmp :
+                                 tmp1 - x*y/(y2 + 0.28f*x2 + FLT_EPSILON);
 
-        cartToPolar = angInDegree == 0 ? cartToPolar : cartToPolar * (float)(180/CV_PI);
+#ifdef DEGREE
+        cartToPolar *= (180/CV_PI);
+#endif
 
-        *((__global float *)((__global char *)dst1 + dst1_index)) = magnitude;
-        *((__global float *)((__global char *)dst2 + dst2_index)) = cartToPolar;
+        dst1[dst1_index] = magnitude;
+        dst2[dst2_index] = cartToPolar;
     }
 }
 
 #if defined (DOUBLE_SUPPORT)
+
 __kernel void arithm_cartToPolar_D6 (__global double *src1, int src1_step, int src1_offset,
                                      __global double *src2, int src2_step, int src2_offset,
                                      __global double *dst1, int dst1_step, int dst1_offset,
                                      __global double *dst2, int dst2_step, int dst2_offset,
-                                     int rows, int cols, int angInDegree)
+                                     int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 3) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 3) + src2_offset);
+        int src1_index = mad24(y, src1_step, x + src1_offset);
+        int src2_index = mad24(y, src2_step, x + src2_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 3) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 3) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        double x = *((__global double *)((__global char *)src1 + src1_index));
-        double y = *((__global double *)((__global char *)src2 + src2_index));
+        double x = src1[src1_index];
+        double y = src2[src2_index];
 
         double x2 = x * x;
         double y2 = y * y;
 
         double magnitude = sqrt(x2 + y2);
-        double cartToPolar;
 
         float tmp = y >= 0 ? 0 : CV_PI*2;
         tmp = x < 0 ? CV_PI : tmp;
 
         float tmp1 = y >= 0 ? CV_PI*0.5 : CV_PI*1.5;
-        cartToPolar = y2 <= x2 ? x*y/(x2 + 0.28f*y2 + (float)DBL_EPSILON)  + tmp :
-                                 tmp1 - x*y/(y2 + 0.28f*x2 + (float)DBL_EPSILON);
+        double cartToPolar = y2 <= x2 ? x*y/(x2 + 0.28f*y2 + DBL_EPSILON)  + tmp :
+                                 tmp1 - x*y/(y2 + 0.28f*x2 + DBL_EPSILON);
 
-        cartToPolar = angInDegree == 0 ? cartToPolar : cartToPolar * (float)(180/CV_PI);
+#ifdef DEGREE
+        cartToPolar *= (180/CV_PI);
+#endif
 
-        *((__global double *)((__global char *)dst1 + dst1_index)) = magnitude;
-        *((__global double *)((__global char *)dst2 + dst2_index)) = cartToPolar;
+        dst1[dst1_index] = magnitude;
+        dst2[dst2_index] = cartToPolar;
     }
 }
+
 #endif

diff --git a/modules/ocl/src/opencl/arithm_compare.cl b/modules/ocl/src/opencl/arithm_compare.cl
index 005d3c7..73e6299 100644 (file)
--- a/modules/ocl/src/opencl/arithm_compare.cl
+++ b/modules/ocl/src/opencl/arithm_compare.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_exp.cl b/modules/ocl/src/opencl/arithm_exp.cl
index 835bc95..f0a1893 100644 (file)
--- a/modules/ocl/src/opencl/arithm_exp.cl
+++ b/modules/ocl/src/opencl/arithm_exp.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_flip.cl b/modules/ocl/src/opencl/arithm_flip.cl
index 416240b..b9bacd3 100644 (file)
--- a/modules/ocl/src/opencl/arithm_flip.cl
+++ b/modules/ocl/src/opencl/arithm_flip.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_log.cl b/modules/ocl/src/opencl/arithm_log.cl
index fe1b304..ba5f32d 100644 (file)
--- a/modules/ocl/src/opencl/arithm_log.cl
+++ b/modules/ocl/src/opencl/arithm_log.cl
@@ -43,9 +43,13 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
 //////////////////////////////////////////////////////////////////////////////////////////////////////
 /////////////////////////////////////////////LOG/////////////////////////////////////////////////////

diff --git a/modules/ocl/src/opencl/arithm_magnitude.cl b/modules/ocl/src/opencl/arithm_magnitude.cl
index 7c8cc18..1053efd 100644 (file)
--- a/modules/ocl/src/opencl/arithm_magnitude.cl
+++ b/modules/ocl/src/opencl/arithm_magnitude.cl
@@ -43,54 +43,32 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
-__kernel void arithm_magnitude_D5 (__global float *src1, int src1_step, int src1_offset,
-                                   __global float *src2, int src2_step, int src2_offset,
-                                   __global float *dst,  int dst_step,  int dst_offset,
-                                  int rows, int cols)
-{
-    int x = get_global_id(0);
-    int y = get_global_id(1);
-
-    if (x < cols && y < rows)
-    {
-        int src1_index = mad24(y, src1_step, (x << 2) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 2) + src2_offset);
-        int dst_index  = mad24(y, dst_step,  (x << 2) + dst_offset);
-
-        float data1 = *((__global float *)((__global char *)src1 + src1_index));
-        float data2 = *((__global float *)((__global char *)src2 + src2_index));
-
-        float tmp = sqrt(data1 * data1 + data2 * data2);
-
-        *((__global float *)((__global char *)dst + dst_index)) = tmp;
-    }
-}
-
-#if defined (DOUBLE_SUPPORT)
-__kernel void arithm_magnitude_D6 (__global double *src1, int src1_step, int src1_offset,
-                                   __global double *src2, int src2_step, int src2_offset,
-                                   __global double *dst,  int dst_step,  int dst_offset,
-                                  int rows, int cols)
+__kernel void arithm_magnitude(__global T *src1, int src1_step, int src1_offset,
+                               __global T *src2, int src2_step, int src2_offset,
+                               __global T *dst,  int dst_step,  int dst_offset,
+                               int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 3) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 3) + src2_offset);
-        int dst_index  = mad24(y, dst_step,  (x << 3) + dst_offset);
-
-        double data1 = *((__global double *)((__global char *)src1 + src1_index));
-        double data2 = *((__global double *)((__global char *)src2 + src2_index));
+        int src1_index = mad24(y, src1_step, x + src1_offset);
+        int src2_index = mad24(y, src2_step, x + src2_offset);
+        int dst_index  = mad24(y, dst_step,  x + dst_offset);
 
-        double tmp = sqrt(data1 * data1 + data2 * data2);
+        T data1 = src1[src1_index];
+        T data2 = src2[src2_index];
 
-        *((__global double *)((__global char *)dst + dst_index)) = tmp;
+        T tmp = hypot(data1, data2);
+        dst[dst_index] = tmp;
     }
 }
-#endif

diff --git a/modules/ocl/src/opencl/arithm_minMax.cl b/modules/ocl/src/opencl/arithm_minMax.cl
index 33a39d8..01db7d0 100644 (file)
--- a/modules/ocl/src/opencl/arithm_minMax.cl
+++ b/modules/ocl/src/opencl/arithm_minMax.cl
@@ -45,7 +45,7 @@
 
 /**************************************PUBLICFUNC*************************************/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)

diff --git a/modules/ocl/src/opencl/arithm_minMaxLoc.cl b/modules/ocl/src/opencl/arithm_minMaxLoc.cl
index 076fb06..1aac6c1 100644 (file)
--- a/modules/ocl/src/opencl/arithm_minMaxLoc.cl
+++ b/modules/ocl/src/opencl/arithm_minMaxLoc.cl
@@ -44,8 +44,13 @@
 //M*/
 
 /**************************************PUBLICFUNC*************************************/
-#if defined (DOUBLE_SUPPORT)
+
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
 #define RES_TYPE double4
 #define CONVERT_RES_TYPE convert_double4
 #else
@@ -222,8 +227,9 @@ __kernel void arithm_op_minMaxLoc(int cols, int invalid_cols, int offset, int el
     {
         localmem_min[lid] = min(minval,localmem_min[lid]);
         localmem_max[lid] = max(maxval,localmem_max[lid]);
-        localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == minval, minloc, localmem_minloc[lid]);
-        localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == maxval, maxloc, localmem_maxloc[lid]);
+        VEC_TYPE minVal = localmem_min[lid], maxVal = localmem_max[lid];
+        localmem_minloc[lid] = CONDITION_FUNC(minVal == minval, minloc, localmem_minloc[lid]);
+        localmem_maxloc[lid] = CONDITION_FUNC(maxVal == maxval, maxloc, localmem_maxloc[lid]);
     }
     barrier(CLK_LOCAL_MEM_FENCE);
 
@@ -234,8 +240,10 @@ __kernel void arithm_op_minMaxLoc(int cols, int invalid_cols, int offset, int el
             int lid2 = lsize + lid;
             localmem_min[lid] = min(localmem_min[lid], localmem_min[lid2]);
             localmem_max[lid] = max(localmem_max[lid], localmem_max[lid2]);
-            localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == localmem_min[lid2], localmem_minloc[lid2], localmem_minloc[lid]);
-            localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == localmem_max[lid2], localmem_maxloc[lid2], localmem_maxloc[lid]);
+            VEC_TYPE min1 = localmem_min[lid], min2 = localmem_min[lid2];
+            localmem_minloc[lid] = CONDITION_FUNC(min1 == min2, localmem_minloc[lid2], localmem_minloc[lid]);
+            VEC_TYPE max1 = localmem_max[lid], max2 = localmem_max[lid2];
+            localmem_maxloc[lid] = CONDITION_FUNC(max1 == max2, localmem_maxloc[lid2], localmem_maxloc[lid]);
        }
        barrier(CLK_LOCAL_MEM_FENCE);
     }

diff --git a/modules/ocl/src/opencl/arithm_minMaxLoc_mask.cl b/modules/ocl/src/opencl/arithm_minMaxLoc_mask.cl
index 4d73be9..84b26ca 100644 (file)
--- a/modules/ocl/src/opencl/arithm_minMaxLoc_mask.cl
+++ b/modules/ocl/src/opencl/arithm_minMaxLoc_mask.cl
@@ -44,8 +44,13 @@
 //M*/
 
 /**************************************PUBLICFUNC*************************************/
-#if defined (DOUBLE_SUPPORT)
+
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
 #define RES_TYPE double4
 #define CONVERT_RES_TYPE convert_double4
 #else
@@ -152,24 +157,26 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
     int  id = get_global_id(0);
     int idx = id + (id / cols) * invalid_cols;
     int midx = id + (id / cols) * minvalid_cols;
+
     __local VEC_TYPE lm_max[128],lm_min[128];
-    VEC_TYPE minval,maxval,temp,m_temp;
-    __local VEC_TYPE_LOC lm_maxloc[128],lm_minloc[128];
-    VEC_TYPE_LOC minloc,maxloc,temploc,negative = -1,one = 1,zero = 0;
+    VEC_TYPE minval, maxval, temp, m_temp, zeroVal = (VEC_TYPE)(0);
+    __local VEC_TYPE_LOC lm_maxloc[128], lm_minloc[128];
+    VEC_TYPE_LOC minloc, maxloc, temploc, negative = -1, one = 1, zero = 0;
+
     if(id < elemnum)
     {
         temp = vload4(idx, &src[offset]);
         m_temp = CONVERT_TYPE(vload4(midx,&mask[moffset]));
         int idx_c = (idx << 2) + offset;
         temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
-        if(id % cols == cols - 1)
+        if (id % cols == cols - 1)
         {
             repeat_me(m_temp);
             repeat_e(temploc);
         }
-        minval = m_temp != (VEC_TYPE)0 ? temp : (VEC_TYPE)MAX_VAL;
-        maxval = m_temp != (VEC_TYPE)0 ? temp : (VEC_TYPE)MIN_VAL;
-        minloc = CONDITION_FUNC(m_temp != (VEC_TYPE)0, temploc , negative);
+        minval = m_temp != zeroVal ? temp : (VEC_TYPE)MAX_VAL;
+        maxval = m_temp != zeroVal ? temp : (VEC_TYPE)MIN_VAL;
+        minloc = CONDITION_FUNC(m_temp != zeroVal, temploc , negative);
         maxloc = minloc;
     }
     else
@@ -179,6 +186,7 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
         minloc = negative;
         maxloc = negative;
     }
+
     for(id=id + (groupnum << 8); id < elemnum;id = id + (groupnum << 8))
     {
         idx = id + (id / cols) * invalid_cols;
@@ -187,17 +195,18 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
         m_temp = CONVERT_TYPE(vload4(midx,&mask[moffset]));
         int idx_c = (idx << 2) + offset;
         temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
-        if(id % cols == cols - 1)
+        if (id % cols == cols - 1)
         {
             repeat_me(m_temp);
             repeat_e(temploc);
         }
-        minval = min(minval,m_temp != (VEC_TYPE)0 ? temp : minval);
-        maxval = max(maxval,m_temp != (VEC_TYPE)0 ? temp : maxval);
+        minval = min(minval, m_temp != zeroVal ? temp : minval);
+        maxval = max(maxval, m_temp != zeroVal ? temp : maxval);
 
-        minloc = CONDITION_FUNC((minval == temp) && (m_temp != (VEC_TYPE)0), temploc , minloc);
-        maxloc = CONDITION_FUNC((maxval == temp) && (m_temp != (VEC_TYPE)0), temploc , maxloc);
+        minloc = CONDITION_FUNC(minval == temp && m_temp != zeroVal, temploc , minloc);
+        maxloc = CONDITION_FUNC(maxval == temp && m_temp != zeroVal, temploc , maxloc);
     }
+
     if(lid > 127)
     {
         lm_min[lid - 128] = minval;
@@ -206,32 +215,37 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
         lm_maxloc[lid - 128] = maxloc;
     }
     barrier(CLK_LOCAL_MEM_FENCE);
+
     if(lid < 128)
     {
-        lm_min[lid] = min(minval,lm_min[lid]);
-        lm_max[lid] = max(maxval,lm_max[lid]);
+        lm_min[lid] = min(minval, lm_min[lid]);
+        lm_max[lid] = max(maxval, lm_max[lid]);
         VEC_TYPE con_min = CONVERT_TYPE(minloc != negative ? one : zero);
         VEC_TYPE con_max = CONVERT_TYPE(maxloc != negative ? one : zero);
-        lm_minloc[lid] = CONDITION_FUNC((lm_min[lid] == minval) && (con_min != (VEC_TYPE)0), minloc , lm_minloc[lid]);
-        lm_maxloc[lid] = CONDITION_FUNC((lm_max[lid] == maxval) && (con_max != (VEC_TYPE)0), maxloc , lm_maxloc[lid]);
+        VEC_TYPE lmMinVal = lm_min[lid], lmMaxVal = lm_max[lid];
+        lm_minloc[lid] = CONDITION_FUNC(lmMinVal == minval && con_min != zeroVal, minloc , lm_minloc[lid]);
+        lm_maxloc[lid] = CONDITION_FUNC(lmMaxVal == maxval && con_max != zeroVal, maxloc , lm_maxloc[lid]);
     }
     barrier(CLK_LOCAL_MEM_FENCE);
+
     for(int lsize = 64; lsize > 0; lsize >>= 1)
     {
         if(lid < lsize)
         {
             int lid2 = lsize + lid;
-            lm_min[lid] = min(lm_min[lid] , lm_min[lid2]);
-            lm_max[lid] = max(lm_max[lid] , lm_max[lid2]);
+            lm_min[lid] = min(lm_min[lid], lm_min[lid2]);
+            lm_max[lid] = max(lm_max[lid], lm_max[lid2]);
             VEC_TYPE con_min = CONVERT_TYPE(lm_minloc[lid2] != negative ? one : zero);
             VEC_TYPE con_max = CONVERT_TYPE(lm_maxloc[lid2] != negative ? one : zero);
-            lm_minloc[lid] =
-                CONDITION_FUNC((lm_min[lid] == lm_min[lid2]) && (con_min != (VEC_TYPE)0), lm_minloc[lid2] , lm_minloc[lid]);
-            lm_maxloc[lid] =
-                CONDITION_FUNC((lm_max[lid] == lm_max[lid2]) && (con_max != (VEC_TYPE)0), lm_maxloc[lid2] , lm_maxloc[lid]);
+
+            VEC_TYPE lmMinVal1 = lm_min[lid], lmMinVal2 = lm_min[lid2];
+            VEC_TYPE lmMaxVal1 = lm_max[lid], lmMaxVal2 = lm_max[lid2];
+            lm_minloc[lid] = CONDITION_FUNC(lmMinVal1 == lmMinVal2 && con_min != zeroVal, lm_minloc[lid2] , lm_minloc[lid]);
+            lm_maxloc[lid] = CONDITION_FUNC(lmMaxVal1 == lmMaxVal2 && con_max != zeroVal, lm_maxloc[lid2] , lm_maxloc[lid]);
         }
         barrier(CLK_LOCAL_MEM_FENCE);
     }
+
     if( lid == 0)
     {
         dst[gid] = CONVERT_RES_TYPE(lm_min[0]);

diff --git a/modules/ocl/src/opencl/arithm_nonzero.cl b/modules/ocl/src/opencl/arithm_nonzero.cl
index fc98257..3180c26 100644 (file)
--- a/modules/ocl/src/opencl/arithm_nonzero.cl
+++ b/modules/ocl/src/opencl/arithm_nonzero.cl
@@ -42,7 +42,7 @@
 // the use of this software, even if advised of the possibility of such damage.
 //
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)

diff --git a/modules/ocl/src/opencl/arithm_phase.cl b/modules/ocl/src/opencl/arithm_phase.cl
index f983594..40346b2 100644 (file)
--- a/modules/ocl/src/opencl/arithm_phase.cl
+++ b/modules/ocl/src/opencl/arithm_phase.cl
@@ -44,17 +44,17 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
-    #ifdef cl_khr_fp64
-        #pragma OPENCL EXTENSION cl_khr_fp64:enable
-    #elif defined (cl_amd_fp64)
-        #pragma OPENCL EXTENSION cl_amd_fp64:enable
-    #endif
-    #define CV_PI 3.1415926535897932384626433832795
-    #define CV_2PI 2*CV_PI
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
+#define CV_PI M_PI
+#define CV_2PI (2 * CV_PI)
 #else
-    #define CV_PI 3.1415926535897932384626433832795f
-    #define CV_2PI 2*CV_PI
+#define CV_PI M_PI_F
+#define CV_2PI (2 * CV_PI)
 #endif
 
 /**************************************phase inradians**************************************/
@@ -159,7 +159,7 @@ __kernel void arithm_phase_indegrees_D6 (__global double *src1, int src1_step1,
 
         double data1 = src1[src1_index];
         double data2 = src2[src2_index];
-        double tmp = atan2(src2[src2_index], src1[src1_index]);
+        double tmp = atan2(data2, data1);
 
         tmp = 180 * tmp / CV_PI;
         if (tmp < 0)

diff --git a/modules/ocl/src/opencl/arithm_polarToCart.cl b/modules/ocl/src/opencl/arithm_polarToCart.cl
index 8469cdb..024f1f0 100644 (file)
--- a/modules/ocl/src/opencl/arithm_polarToCart.cl
+++ b/modules/ocl/src/opencl/arithm_polarToCart.cl
@@ -44,42 +44,51 @@
 //M*/
 
 #ifdef DOUBLE_SUPPORT
-    #pragma OPENCL EXTENSION cl_khr_fp64:enable
-    #define CV_PI   3.1415926535897932384626433832795
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
+#define CV_PI M_PI
 #else
-    #define CV_PI   3.1415926535897932384626433832795f
+#define CV_PI M_PI_F
 #endif
 
 /////////////////////////////////////////////////////////////////////////////////////////////////////
 /////////////////////////////////////////polarToCart with magnitude//////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void arithm_polarToCart_mag_D5 (__global float *src1, int src1_step, int src1_offset,//magnitue
                                          __global float *src2, int src2_step, int src2_offset,//angle
                                          __global float *dst1, int dst1_step, int dst1_offset,
                                          __global float *dst2, int dst2_step, int dst2_offset,
-                                         int rows, int cols, int angInDegree)
+                                         int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 2) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 2) + src2_offset);
+        int src1_index = mad24(y, src1_step, x + src1_offset);
+        int src2_index = mad24(y, src2_step, x + src2_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 2) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 2) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        float x = *((__global float *)((__global char *)src1 + src1_index));
-        float y = *((__global float *)((__global char *)src2 + src2_index));
+        float x = src1[src1_index];
+        float y = src2[src2_index];
 
+#ifdef DEGREE
         float ascale = CV_PI/180.0f;
-        float alpha  = angInDegree == 1 ? y * ascale : y;
+        float alpha = y * ascale;
+#else
+        float alpha = y;
+#endif
         float a = cos(alpha) * x;
         float b = sin(alpha) * x;
 
-        *((__global float *)((__global char *)dst1 + dst1_index)) = a;
-        *((__global float *)((__global char *)dst2 + dst2_index)) = b;
+        dst1[dst1_index] = a;
+        dst2[dst2_index] = b;
     }
 }
 
@@ -88,29 +97,33 @@ __kernel void arithm_polarToCart_mag_D6 (__global double *src1, int src1_step, i
                                          __global double *src2, int src2_step, int src2_offset,//angle
                                          __global double *dst1, int dst1_step, int dst1_offset,
                                          __global double *dst2, int dst2_step, int dst2_offset,
-                                         int rows, int cols, int angInDegree)
+                                         int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 3) + src1_offset);
-        int src2_index = mad24(y, src2_step, (x << 3) + src2_offset);
+        int src1_index = mad24(y, src1_step, x + src1_offset);
+        int src2_index = mad24(y, src2_step, x + src2_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 3) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 3) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        double x = *((__global double *)((__global char *)src1 + src1_index));
-        double y = *((__global double *)((__global char *)src2 + src2_index));
+        double x = src1[src1_index];
+        double y = src2[src2_index];
 
+#ifdef DEGREE
         float ascale = CV_PI/180.0;
-        double alpha  = angInDegree == 1 ? y * ascale : y;
+        float alpha = y * ascale;
+#else
+        float alpha = y;
+#endif
         double a = cos(alpha) * x;
         double b = sin(alpha) * x;
 
-        *((__global double *)((__global char *)dst1 + dst1_index)) = a;
-        *((__global double *)((__global char *)dst2 + dst2_index)) = b;
+        dst1[dst1_index] = a;
+        dst2[dst2_index] = b;
     }
 }
 #endif
@@ -118,30 +131,35 @@ __kernel void arithm_polarToCart_mag_D6 (__global double *src1, int src1_step, i
 /////////////////////////////////////////////////////////////////////////////////////////////////////
 /////////////////////////////////////////polarToCart without magnitude//////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void arithm_polarToCart_D5 (__global float *src,  int src_step,  int src_offset,//angle
                                      __global float *dst1, int dst1_step, int dst1_offset,
                                      __global float *dst2, int dst2_step, int dst2_offset,
-                                     int rows, int cols, int angInDegree)
+                                     int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src_index  = mad24(y, src_step,  (x << 2) + src_offset);
+        int src_index  = mad24(y, src_step,  x + src_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 2) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 2) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        float y = *((__global float *)((__global char *)src + src_index));
+        float y = src[src_index];
 
+#ifdef DEGREE
         float ascale = CV_PI/180.0f;
-        float alpha  = angInDegree == 1 ? y * ascale : y;
+        float alpha = y * ascale;
+#else
+        float alpha = y;
+#endif
         float a = cos(alpha);
         float b = sin(alpha);
 
-        *((__global float *)((__global char *)dst1 + dst1_index)) = a;
-        *((__global float *)((__global char *)dst2 + dst2_index)) = b;
+        dst1[dst1_index] = a;
+        dst2[dst2_index] = b;
     }
 }
 
@@ -149,27 +167,31 @@ __kernel void arithm_polarToCart_D5 (__global float *src,  int src_step,  int sr
 __kernel void arithm_polarToCart_D6 (__global float *src,  int src_step,  int src_offset,//angle
                                      __global float *dst1, int dst1_step, int dst1_offset,
                                      __global float *dst2, int dst2_step, int dst2_offset,
-                                     int rows, int cols, int angInDegree)
+                                     int rows, int cols)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
 
     if (x < cols && y < rows)
     {
-        int src_index  = mad24(y, src_step,  (x << 3) + src_offset);
+        int src_index  = mad24(y, src_step,  x + src_offset);
 
-        int dst1_index = mad24(y, dst1_step, (x << 3) + dst1_offset);
-        int dst2_index = mad24(y, dst2_step, (x << 3) + dst2_offset);
+        int dst1_index = mad24(y, dst1_step, x + dst1_offset);
+        int dst2_index = mad24(y, dst2_step, x + dst2_offset);
 
-        double y = *((__global double *)((__global char *)src + src_index));
+        double y = src[src_index];
 
-        float ascale = CV_PI/180.0;
-        double alpha  = angInDegree == 1 ? y * ascale : y;
+#ifdef DEGREE
+        float ascale = CV_PI/180.0f;
+        float alpha = y * ascale;
+#else
+        float alpha = y;
+#endif
         double a = cos(alpha);
         double b = sin(alpha);
 
-        *((__global double *)((__global char *)dst1 + dst1_index)) = a;
-        *((__global double *)((__global char *)dst2 + dst2_index)) = b;
+        dst1[dst1_index] = a;
+        dst2[dst2_index] = b;
     }
 }
 #endif

diff --git a/modules/ocl/src/opencl/arithm_pow.cl b/modules/ocl/src/opencl/arithm_pow.cl
index 1704f6b..385e4cc 100644 (file)
--- a/modules/ocl/src/opencl/arithm_pow.cl
+++ b/modules/ocl/src/opencl/arithm_pow.cl
@@ -43,58 +43,31 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
-typedef double F;
-typedef double4 F4;
-#define convert_F4 convert_double4;
-#else
-typedef float F;
-typedef float4 F4;
-#define convert_F4 convert_float4;
 #endif
-/************************************** pow **************************************/
-__kernel void arithm_pow_D5 (__global float *src1, int src1_step, int src1_offset,
-                             __global float *dst,  int dst_step,  int dst_offset,
-                             int rows, int cols, int dst_step1,
-                             F p)
-{
-
-    int x = get_global_id(0);
-    int y = get_global_id(1);
-
-    if(x < cols && y < rows)
-    {
-        int src1_index = mad24(y, src1_step, (x << 2) + src1_offset);
-        int dst_index  = mad24(y, dst_step,  (x << 2) + dst_offset);
-
-        float src1_data = *((__global float *)((__global char *)src1 + src1_index));
-        float tmp = src1_data > 0 ? exp(p * log(src1_data)) : (src1_data == 0 ? 0 : exp(p * log(fabs(src1_data))));
-
-        *((__global float *)((__global char *)dst + dst_index)) = tmp;
-    }
+#endif
 
-}
+/************************************** pow **************************************/
 
-#if defined (DOUBLE_SUPPORT)
-__kernel void arithm_pow_D6 (__global double *src1, int src1_step, int src1_offset,
-                             __global double *dst,  int dst_step,  int dst_offset,
-                             int rows, int cols, int dst_step1,
-                             F p)
+__kernel void arithm_pow(__global VT * src, int src_step, int src_offset,
+                         __global VT * dst, int dst_step, int dst_offset,
+                         int rows, int cols, T p)
 {
-
     int x = get_global_id(0);
     int y = get_global_id(1);
 
-    if(x < cols && y < rows)
+    if (x < cols && y < rows)
     {
-        int src1_index = mad24(y, src1_step, (x << 3) + src1_offset);
-        int dst_index  = mad24(y, dst_step,  (x << 3) + dst_offset);
+        int src_index = mad24(y, src_step, x + src_offset);
+        int dst_index = mad24(y, dst_step, x + dst_offset);
 
-        double src1_data = *((__global double *)((__global char *)src1 + src1_index));
-        double tmp = src1_data > 0 ? exp(p * log(src1_data)) : (src1_data == 0 ? 0 : exp(p * log(fabs(src1_data))));
-        *((__global double *)((__global char *)dst + dst_index)) = tmp;
-    }
+        VT src_data = src[src_index];
+        VT tmp = src_data > 0 ? exp(p * log(src_data)) : (src_data == 0 ? 0 : exp(p * log(fabs(src_data))));
 
+        dst[dst_index] = tmp;
+    }
 }
-#endif

diff --git a/modules/ocl/src/opencl/arithm_setidentity.cl b/modules/ocl/src/opencl/arithm_setidentity.cl
index fb684c3..0ead5b0 100644 (file)
--- a/modules/ocl/src/opencl/arithm_setidentity.cl
+++ b/modules/ocl/src/opencl/arithm_setidentity.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_sum.cl b/modules/ocl/src/opencl/arithm_sum.cl
index 7ada5be..514cf2a 100644 (file)
--- a/modules/ocl/src/opencl/arithm_sum.cl
+++ b/modules/ocl/src/opencl/arithm_sum.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/arithm_transpose.cl b/modules/ocl/src/opencl/arithm_transpose.cl
index bd06a52..8cde654 100644 (file)
--- a/modules/ocl/src/opencl/arithm_transpose.cl
+++ b/modules/ocl/src/opencl/arithm_transpose.cl
@@ -43,7 +43,7 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)

diff --git a/modules/ocl/src/opencl/bgfg_mog.cl b/modules/ocl/src/opencl/bgfg_mog.cl
index a13a30e..06e18c2 100644 (file)
--- a/modules/ocl/src/opencl/bgfg_mog.cl
+++ b/modules/ocl/src/opencl/bgfg_mog.cl
@@ -67,11 +67,14 @@ static float clamp1(float var, float learningRate, float diff, float minVar)
 {
     return fmax(var + learningRate * (diff * diff - var), minVar);
 }
+
 #else
+
 #define T_FRAME uchar4
 #define T_MEAN_VAR float4
 #define CONVERT_TYPE convert_uchar4_sat
 #define F_ZERO (0.0f, 0.0f, 0.0f, 0.0f)
+
 inline float4 cvt(const uchar4 val)
 {
     float4 result;
@@ -93,6 +96,14 @@ inline float sum(const float4 val)
     return (val.x + val.y + val.z);
 }
 
+static void swap4(__global float4* ptr, int x, int y, int k, int rows, int ptr_step)
+{
+    float4 val = ptr[(k * rows + y) * ptr_step + x];
+    ptr[(k * rows + y) * ptr_step + x] = ptr[((k + 1) * rows + y) * ptr_step + x];
+    ptr[((k + 1) * rows + y) * ptr_step + x] = val;
+}
+
+
 static float4 clamp1(const float4 var, float learningRate, const float4 diff, float minVar)
 {
     float4 result;
@@ -102,6 +113,7 @@ static float4 clamp1(const float4 var, float learningRate, const float4 diff, fl
     result.w = 0.0f;
     return result;
 }
+
 #endif
 
 typedef struct
@@ -114,7 +126,7 @@ typedef struct
     float c_varMax;
     float c_tau;
     uchar c_shadowVal;
-}con_srtuct_t;
+} con_srtuct_t;
 
 static void swap(__global float* ptr, int x, int y, int k, int rows, int ptr_step)
 {
@@ -123,13 +135,6 @@ static void swap(__global float* ptr, int x, int y, int k, int rows, int ptr_ste
     ptr[((k + 1) * rows + y) * ptr_step + x] = val;
 }
 
-static void swap4(__global float4* ptr, int x, int y, int k, int rows, int ptr_step)
-{
-    float4 val = ptr[(k * rows + y) * ptr_step + x];
-    ptr[(k * rows + y) * ptr_step + x] = ptr[((k + 1) * rows + y) * ptr_step + x];
-    ptr[((k + 1) * rows + y) * ptr_step + x] = val;
-}
-
 __kernel void mog_withoutLearning_kernel(__global T_FRAME* frame, __global uchar* fgmask,
     __global float* weight, __global T_MEAN_VAR* mean, __global T_MEAN_VAR* var,
     int frame_row, int frame_col, int frame_step, int fgmask_step,

diff --git a/modules/ocl/src/opencl/blend_linear.cl b/modules/ocl/src/opencl/blend_linear.cl
index 06a51f2..bc7aa46 100644 (file)
--- a/modules/ocl/src/opencl/blend_linear.cl
+++ b/modules/ocl/src/opencl/blend_linear.cl
@@ -43,7 +43,7 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)

diff --git a/modules/ocl/src/opencl/brute_force_match.cl b/modules/ocl/src/opencl/brute_force_match.cl
index ce0d86e..a005284 100644 (file)
--- a/modules/ocl/src/opencl/brute_force_match.cl
+++ b/modules/ocl/src/opencl/brute_force_match.cl
@@ -63,14 +63,6 @@
 #define DIST_TYPE 0
 #endif
 
-//http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallel
-static int bit1Count(int v)
-{
-    v = v - ((v >> 1) & 0x55555555);                    // reuse input as temporary
-    v = (v & 0x33333333) + ((v >> 2) & 0x33333333);     // temp
-    return ((v + (v >> 4) & 0xF0F0F0F) * 0x1010101) >> 24; // count
-}
-
 // dirty fix for non-template support
 #if   (DIST_TYPE == 0) // L1Dist
 #   ifdef T_FLOAT
@@ -89,6 +81,13 @@ typedef float value_type;
 typedef float result_type;
 #define DIST_RES(x) sqrt(x)
 #elif (DIST_TYPE == 2) // Hamming
+//http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallel
+static int bit1Count(int v)
+{
+    v = v - ((v >> 1) & 0x55555555);                    // reuse input as temporary
+    v = (v & 0x33333333) + ((v >> 2) & 0x33333333);     // temp
+    return ((v + (v >> 4) & 0xF0F0F0F) * 0x1010101) >> 24; // count
+}
 #define DIST(x, y) bit1Count( (x) ^ (y) )
 typedef int value_type;
 typedef int result_type;

diff --git a/modules/ocl/src/opencl/build_warps.cl b/modules/ocl/src/opencl/build_warps.cl
index 4402e8c..bd5e002 100644 (file)
--- a/modules/ocl/src/opencl/build_warps.cl
+++ b/modules/ocl/src/opencl/build_warps.cl
@@ -43,31 +43,25 @@
 //
 //M*/
 
-__kernel
-    void buildWarpPlaneMaps
-    (
-    __global float * map_x,
-    __global float * map_y,
-    __constant float * KRT,
-    int tl_u,
-    int tl_v,
-    int cols,
-    int rows,
-    int step_x,
-    int step_y,
-    float scale
-    )
+__kernel void buildWarpPlaneMaps(__global float * xmap, __global float * ymap,
+                                 __constant float * KRT,
+                                 int tl_u, int tl_v,
+                                 int cols, int rows,
+                                 int xmap_step, int ymap_step,
+                                 int xmap_offset, int ymap_offset,
+                                 float scale)
 {
     int du = get_global_id(0);
     int dv = get_global_id(1);
-    step_x /= sizeof(float);
-    step_y /= sizeof(float);
 
     __constant float * ck_rinv = KRT;
     __constant float * ct      = KRT + 9;
 
     if (du < cols && dv < rows)
     {
+        int xmap_index = mad24(dv, xmap_step, xmap_offset + du);
+        int ymap_index = mad24(dv, ymap_step, ymap_offset + du);
+
         float u = tl_u + du;
         float v = tl_v + dv;
         float x, y;
@@ -83,33 +77,27 @@ __kernel
         x /= z;
         y /= z;
 
-        map_x[dv * step_x + du] = x;
-        map_y[dv * step_y + du] = y;
+        xmap[xmap_index] = x;
+        ymap[ymap_index] = y;
     }
 }
 
-__kernel
-    void buildWarpCylindricalMaps
-    (
-    __global float * map_x,
-    __global float * map_y,
-    __constant float * ck_rinv,
-    int tl_u,
-    int tl_v,
-    int cols,
-    int rows,
-    int step_x,
-    int step_y,
-    float scale
-    )
+__kernel void buildWarpCylindricalMaps(__global float * xmap, __global float * ymap,
+                                       __constant float * ck_rinv,
+                                       int tl_u, int tl_v,
+                                       int cols, int rows,
+                                       int xmap_step, int ymap_step,
+                                       int xmap_offset, int ymap_offset,
+                                       float scale)
 {
     int du = get_global_id(0);
     int dv = get_global_id(1);
-    step_x /= sizeof(float);
-    step_y /= sizeof(float);
 
     if (du < cols && dv < rows)
     {
+        int xmap_index = mad24(dv, xmap_step, xmap_offset + du);
+        int ymap_index = mad24(dv, ymap_step, ymap_offset + du);
+
         float u = tl_u + du;
         float v = tl_v + dv;
         float x, y;
@@ -127,33 +115,27 @@ __kernel
         if (z > 0) { x /= z; y /= z; }
         else x = y = -1;
 
-        map_x[dv * step_x + du] = x;
-        map_y[dv * step_y + du] = y;
+        xmap[xmap_index] = x;
+        ymap[ymap_index] = y;
     }
 }
 
-__kernel
-    void buildWarpSphericalMaps
-    (
-    __global float * map_x,
-    __global float * map_y,
-    __constant float * ck_rinv,
-    int tl_u,
-    int tl_v,
-    int cols,
-    int rows,
-    int step_x,
-    int step_y,
-    float scale
-    )
+__kernel void buildWarpSphericalMaps(__global float * xmap, __global float * ymap,
+                                     __constant float * ck_rinv,
+                                     int tl_u, int tl_v,
+                                     int cols, int rows,
+                                     int xmap_step, int ymap_step,
+                                     int xmap_offset, int ymap_offset,
+                                     float scale)
 {
     int du = get_global_id(0);
     int dv = get_global_id(1);
-    step_x /= sizeof(float);
-    step_y /= sizeof(float);
 
     if (du < cols && dv < rows)
     {
+        int xmap_index = mad24(dv, xmap_step, xmap_offset + du);
+        int ymap_index = mad24(dv, ymap_step, ymap_offset + du);
+
         float u = tl_u + du;
         float v = tl_v + dv;
         float x, y;
@@ -174,63 +156,52 @@ __kernel
         if (z > 0) { x /= z; y /= z; }
         else x = y = -1;
 
-        map_x[dv * step_x + du] = x;
-        map_y[dv * step_y + du] = y;
+        xmap[xmap_index] = x;
+        ymap[ymap_index] = y;
     }
 }
 
-__kernel
-    void buildWarpAffineMaps
-    (
-    __global float * xmap,
-    __global float * ymap,
-    __constant float * c_warpMat,
-    int cols,
-    int rows,
-    int step_x,
-    int step_y
-    )
+__kernel void buildWarpAffineMaps(__global float * xmap, __global float * ymap,
+                                  __constant float * c_warpMat,
+                                  int cols, int rows,
+                                  int xmap_step, int ymap_step,
+                                  int xmap_offset, int ymap_offset)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
-    step_x /= sizeof(float);
-    step_y /= sizeof(float);
 
     if (x < cols && y < rows)
     {
-        const float xcoo = c_warpMat[0] * x + c_warpMat[1] * y + c_warpMat[2];
-        const float ycoo = c_warpMat[3] * x + c_warpMat[4] * y + c_warpMat[5];
+        int xmap_index = mad24(y, xmap_step, x + xmap_offset);
+        int ymap_index = mad24(y, ymap_step, x + ymap_offset);
+
+        float xcoo = c_warpMat[0] * x + c_warpMat[1] * y + c_warpMat[2];
+        float ycoo = c_warpMat[3] * x + c_warpMat[4] * y + c_warpMat[5];
 
-        map_x[y * step_x + x] = xcoo;
-        map_y[y * step_y + x] = ycoo;
+        xmap[xmap_index] = xcoo;
+        ymap[ymap_index] = ycoo;
     }
 }
 
-__kernel
-    void buildWarpPerspectiveMaps
-    (
-    __global float * xmap,
-    __global float * ymap,
-    __constant float * c_warpMat,
-    int cols,
-    int rows,
-    int step_x,
-    int step_y
-    )
+__kernel void buildWarpPerspectiveMaps(__global float * xmap, __global float * ymap,
+                                       __constant float * c_warpMat,
+                                       int cols, int rows,
+                                       int xmap_step, int ymap_step,
+                                       int xmap_offset, int ymap_offset)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
-    step_x /= sizeof(float);
-    step_y /= sizeof(float);
 
     if (x < cols && y < rows)
     {
-        const float coeff = 1.0f / (c_warpMat[6] * x + c_warpMat[7] * y + c_warpMat[8]);
+        int xmap_index = mad24(y, xmap_step, x + xmap_offset);
+        int ymap_index = mad24(y, ymap_step, x + ymap_offset);
 
-        const float xcoo = coeff * (c_warpMat[0] * x + c_warpMat[1] * y + c_warpMat[2]);
-        const float ycoo = coeff * (c_warpMat[3] * x + c_warpMat[4] * y + c_warpMat[5]);
+        float coeff = 1.0f / (c_warpMat[6] * x + c_warpMat[7] * y + c_warpMat[8]);
+        float xcoo = coeff * (c_warpMat[0] * x + c_warpMat[1] * y + c_warpMat[2]);
+        float ycoo = coeff * (c_warpMat[3] * x + c_warpMat[4] * y + c_warpMat[5]);
 
-        map_x[y * step_x + x] = xcoo;
-        map_y[y * step_y + x] = ycoo;
+        xmap[xmap_index] = xcoo;
+        ymap[ymap_index] = ycoo;
     }
 }

diff --git a/modules/ocl/src/opencl/convertC3C4.cl b/modules/ocl/src/opencl/convertC3C4.cl
index b3e699d..4c519fd 100644 (file)
--- a/modules/ocl/src/opencl/convertC3C4.cl
+++ b/modules/ocl/src/opencl/convertC3C4.cl
@@ -33,12 +33,17 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
-__kernel void convertC3C4(__global const GENTYPE4 * restrict src, __global GENTYPE4 *dst, int cols, int rows,
-                    int dstStep_in_piexl,int pixel_end)
+__kernel void convertC3C4(__global const GENTYPE4 * restrict src, __global GENTYPE4 *dst,
+                         int cols, int rows,
+                         int dstStep_in_piexl, int pixel_end)
 {
     int id = get_global_id(0);
     int3 pixelid = (int3)(mul24(id,3),mad24(id,3,1),mad24(id,3,2));
@@ -88,13 +93,12 @@ __kernel void convertC3C4(__global const GENTYPE4 * restrict src, __global GENTY
         dst[addr.y] = outpix1;
     }
     else if(outx.x<cols && outy.x<rows)
-    {
         dst[addr.x] = outpix0;
-    }
 }
 
-__kernel void convertC4C3(__global const GENTYPE4 * restrict src, __global GENTYPE4 *dst, int cols, int rows,
-                    int srcStep_in_pixel,int pixel_end)
+__kernel void convertC4C3(__global const GENTYPE4 * restrict src, __global GENTYPE4 *dst,
+                          int cols, int rows,
+                          int srcStep_in_pixel, int pixel_end)
 {
     int id = get_global_id(0)<<2;
     int y = id / cols;
@@ -145,7 +149,5 @@ __kernel void convertC4C3(__global const GENTYPE4 * restrict src, __global GENTY
         dst[outaddr.y] = outpixel1;
     }
     else if(outaddr.x <= pixel_end)
-    {
         dst[outaddr.x] = pixel0;
-    }
 }

diff --git a/modules/ocl/src/opencl/cvt_color.cl b/modules/ocl/src/opencl/cvt_color.cl
index 01286f7..bf3b6cf 100644 (file)
--- a/modules/ocl/src/opencl/cvt_color.cl
+++ b/modules/ocl/src/opencl/cvt_color.cl
@@ -46,41 +46,45 @@
 
 /**************************************PUBLICFUNC*************************************/
 
-#if defined (DOUBLE_SUPPORT)
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
+#ifndef hscale
+#define hscale 0
 #endif
 
-#if defined (DEPTH_0)
+#ifndef hrange
+#define hrange 0
+#endif
+
+#ifdef DEPTH_0
 #define DATA_TYPE uchar
+#define COEFF_TYPE int
 #define MAX_NUM  255
 #define HALF_MAX 128
-#define SAT_CAST(num) convert_uchar_sat(num)
+#define SAT_CAST(num) convert_uchar_sat_rte(num)
 #endif
 
-#if defined (DEPTH_2)
+#ifdef DEPTH_2
 #define DATA_TYPE ushort
+#define COEFF_TYPE int
 #define MAX_NUM  65535
 #define HALF_MAX 32768
-#define SAT_CAST(num) convert_ushort_sat(num)
+#define SAT_CAST(num) convert_ushort_sat_rte(num)
 #endif
 
-#if defined (DEPTH_5)
+#ifdef DEPTH_5
 #define DATA_TYPE float
+#define COEFF_TYPE float
 #define MAX_NUM  1.0f
 #define HALF_MAX 0.5f
 #define SAT_CAST(num) (num)
 #endif
 
-#ifndef DATA_TYPE
-    #define DATA_TYPE UNDEFINED
-#endif
-
-#define CV_DESCALE(x,n) (((x) + (1 << ((n)-1))) >> (n))
+#define CV_DESCALE(x, n) (((x) + (1 << ((n)-1))) >> (n))
 
 enum
 {
     yuv_shift  = 14,
     xyz_shift  = 12,
+    hsv_shift = 12,
     R2Y        = 4899,
     G2Y        = 9617,
     B2Y        = 1868,
@@ -89,18 +93,18 @@ enum
 
 ///////////////////////////////////// RGB <-> GRAY //////////////////////////////////////
 
-__kernel void RGB2Gray(int cols, int rows, int src_step, int dst_step, int channels,
+__kernel void RGB2Gray(int cols, int rows, int src_step, int dst_step,
                        int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
                        int src_offset, int dst_offset)
 {
-    const int x = get_global_id(0);
-    const int y = get_global_id(1);
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
     if (y < rows && x < cols)
     {
-        int src_idx = mad24(y, src_step, src_offset + x * channels);
+        int src_idx = mad24(y, src_step, src_offset + (x << 2));
         int dst_idx = mad24(y, dst_step, dst_offset + x);
-#if defined (DEPTH_5)
+#ifdef DEPTH_5
         dst[dst_idx] = src[src_idx + bidx] * 0.114f + src[src_idx + 1] * 0.587f + src[src_idx + (bidx^2)] * 0.299f;
 #else
         dst[dst_idx] = (DATA_TYPE)CV_DESCALE((src[src_idx + bidx] * B2Y + src[src_idx + 1] * G2Y + src[src_idx + (bidx^2)] * R2Y), yuv_shift);
@@ -108,22 +112,25 @@ __kernel void RGB2Gray(int cols, int rows, int src_step, int dst_step, int chann
     }
 }
 
-__kernel void Gray2RGB(int cols,int rows,int src_step,int dst_step,
+__kernel void Gray2RGB(int cols, int rows, int src_step, int dst_step, int bidx,
                        __global const DATA_TYPE* src, __global DATA_TYPE* dst,
                        int src_offset, int dst_offset)
 {
-    const int x = get_global_id(0);
-    const int y = get_global_id(1);
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
     if (y < rows && x < cols)
     {
         int src_idx = mad24(y, src_step, src_offset + x);
-        int dst_idx = mad24(y, dst_step, dst_offset + x * 4);
+        int dst_idx = mad24(y, dst_step, dst_offset + (x << 2));
+
         DATA_TYPE val = src[src_idx];
-        dst[dst_idx++] = val;
-        dst[dst_idx++] = val;
-        dst[dst_idx++] = val;
-        dst[dst_idx] = MAX_NUM;
+        dst[dst_idx] = val;
+        dst[dst_idx + 1] = val;
+        dst[dst_idx + 2] = val;
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
     }
 }
 
@@ -132,7 +139,7 @@ __kernel void Gray2RGB(int cols,int rows,int src_step,int dst_step,
 __constant float c_RGB2YUVCoeffs_f[5]  = { 0.114f, 0.587f, 0.299f, 0.492f, 0.877f };
 __constant int   c_RGB2YUVCoeffs_i[5]  = { B2Y, G2Y, R2Y, 8061, 14369 };
 
-__kernel void RGB2YUV(int cols,int rows,int src_step,int dst_step,int channels,
+__kernel void RGB2YUV(int cols, int rows, int src_step, int dst_step,
                       int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
                       int src_offset, int dst_offset)
 {
@@ -141,35 +148,34 @@ __kernel void RGB2YUV(int cols,int rows,int src_step,int dst_step,int channels,
 
     if (y < rows && x < cols)
     {
-        x *= channels;
+        x <<= 2;
         int src_idx = mad24(y, src_step, src_offset + x);
         int dst_idx = mad24(y, dst_step, dst_offset + x);
-        dst += dst_idx;
-        const DATA_TYPE rgb[] = {src[src_idx], src[src_idx + 1], src[src_idx + 2]};
+        DATA_TYPE rgb[] = { src[src_idx], src[src_idx + 1], src[src_idx + 2] };
 
-#if defined (DEPTH_5)
+#ifdef DEPTH_5
         __constant float * coeffs = c_RGB2YUVCoeffs_f;
-        const DATA_TYPE Y  = rgb[0] * coeffs[bidx] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx^2];
-        const DATA_TYPE Cr = (rgb[bidx] - Y) * coeffs[3] + HALF_MAX;
-        const DATA_TYPE Cb = (rgb[bidx^2] - Y) * coeffs[4] + HALF_MAX;
+        DATA_TYPE Y  = rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx];
+        DATA_TYPE Cr = (rgb[bidx^2] - Y) * coeffs[3] + HALF_MAX;
+        DATA_TYPE Cb = (rgb[bidx] - Y) * coeffs[4] + HALF_MAX;
 #else
         __constant int * coeffs = c_RGB2YUVCoeffs_i;
-        const int delta = HALF_MAX * (1 << yuv_shift);
-        const int Y =  CV_DESCALE(rgb[0] * coeffs[bidx] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx^2], yuv_shift);
-        const int Cr = CV_DESCALE((rgb[bidx] - Y) * coeffs[3] + delta, yuv_shift);
-        const int Cb = CV_DESCALE((rgb[bidx^2] - Y) * coeffs[4] + delta, yuv_shift);
+        int delta = HALF_MAX * (1 << yuv_shift);
+        int Y =  CV_DESCALE(rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx], yuv_shift);
+        int Cr = CV_DESCALE((rgb[bidx^2] - Y) * coeffs[3] + delta, yuv_shift);
+        int Cb = CV_DESCALE((rgb[bidx] - Y) * coeffs[4] + delta, yuv_shift);
 #endif
 
-        dst[0] = SAT_CAST( Y );
-        dst[1] = SAT_CAST( Cr );
-        dst[2] = SAT_CAST( Cb );
+        dst[dst_idx] = SAT_CAST( Y );
+        dst[dst_idx + 1] = SAT_CAST( Cr );
+        dst[dst_idx + 2] = SAT_CAST( Cb );
     }
 }
 
 __constant float c_YUV2RGBCoeffs_f[5] = { 2.032f, -0.395f, -0.581f, 1.140f };
 __constant int   c_YUV2RGBCoeffs_i[5] = { 33292, -6472, -9519, 18678 };
 
-__kernel void YUV2RGB(int cols,int rows,int src_step,int dst_step,int channels,
+__kernel void YUV2RGB(int cols, int rows, int src_step, int dst_step,
                       int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
                       int src_offset, int dst_offset)
 {
@@ -178,27 +184,29 @@ __kernel void YUV2RGB(int cols,int rows,int src_step,int dst_step,int channels,
 
     if (y < rows && x < cols)
     {
-        x *= channels;
+        x <<= 2;
         int src_idx = mad24(y, src_step, src_offset + x);
         int dst_idx = mad24(y, dst_step, dst_offset + x);
-        dst += dst_idx;
-        const DATA_TYPE yuv[] = {src[src_idx], src[src_idx + 1], src[src_idx + 2]};
+        DATA_TYPE yuv[] = { src[src_idx], src[src_idx + 1], src[src_idx + 2] };
 
-#if defined (DEPTH_5)
+#ifdef DEPTH_5
         __constant float * coeffs = c_YUV2RGBCoeffs_f;
-        const float b = yuv[0] + (yuv[2] - HALF_MAX) * coeffs[3];
-        const float g = yuv[0] + (yuv[2] - HALF_MAX) * coeffs[2] + (yuv[1] - HALF_MAX) * coeffs[1];
-        const float r = yuv[0] + (yuv[1] - HALF_MAX) * coeffs[0];
+        float b = yuv[0] + (yuv[2] - HALF_MAX) * coeffs[3];
+        float g = yuv[0] + (yuv[2] - HALF_MAX) * coeffs[2] + (yuv[1] - HALF_MAX) * coeffs[1];
+        float r = yuv[0] + (yuv[1] - HALF_MAX) * coeffs[0];
 #else
         __constant int * coeffs = c_YUV2RGBCoeffs_i;
-        const int b = yuv[0] + CV_DESCALE((yuv[2] - HALF_MAX) * coeffs[3], yuv_shift);
-        const int g = yuv[0] + CV_DESCALE((yuv[2] - HALF_MAX) * coeffs[2] + (yuv[1] - HALF_MAX) * coeffs[1], yuv_shift);
-        const int r = yuv[0] + CV_DESCALE((yuv[1] - HALF_MAX) * coeffs[0], yuv_shift);
+        int b = yuv[0] + CV_DESCALE((yuv[2] - HALF_MAX) * coeffs[3], yuv_shift);
+        int g = yuv[0] + CV_DESCALE((yuv[2] - HALF_MAX) * coeffs[2] + (yuv[1] - HALF_MAX) * coeffs[1], yuv_shift);
+        int r = yuv[0] + CV_DESCALE((yuv[1] - HALF_MAX) * coeffs[0], yuv_shift);
 #endif
 
-        dst[bidx^2] = SAT_CAST( b );
-        dst[1]      = SAT_CAST( g );
-        dst[bidx]   = SAT_CAST( r );
+        dst[dst_idx + bidx] = SAT_CAST( b );
+        dst[dst_idx + 1]      = SAT_CAST( g );
+        dst[dst_idx + (bidx^2)]   = SAT_CAST( r );
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
     }
 }
 
@@ -209,17 +217,17 @@ __constant int ITUR_BT_601_CVG = 852492;
 __constant int ITUR_BT_601_CVR = 1673527;
 __constant int ITUR_BT_601_SHIFT = 20;
 
-__kernel void YUV2RGBA_NV12(int cols,int rows,int src_step,int dst_step,
-                            int bidx, int width, int height, __global const uchar* src, __global uchar* dst,
+__kernel void YUV2RGBA_NV12(int cols, int rows, int src_step, int dst_step,
+                            int bidx, __global const uchar* src, __global uchar* dst,
                             int src_offset, int dst_offset)
 {
-    const int x = get_global_id(0); // max_x = width / 2
-    const int y = get_global_id(1); // max_y = height/ 2
+    const int x = get_global_id(0);
+    const int y = get_global_id(1);
 
-    if (y < height / 2 && x < width / 2 )
+    if (y < rows / 2 && x < cols / 2 )
     {
         __global const uchar* ysrc = src + mad24(y << 1, src_step, (x << 1) + src_offset);
-        __global const uchar* usrc = src + mad24(height + y, src_step, (x << 1) + src_offset);
+        __global const uchar* usrc = src + mad24(rows + y, src_step, (x << 1) + src_offset);
         __global uchar*       dst1 = dst + mad24(y << 1, dst_step, (x << 3) + dst_offset);
         __global uchar*       dst2 = dst + mad24((y << 1) + 1, dst_step, (x << 3) + dst_offset);
 
@@ -261,12 +269,12 @@ __kernel void YUV2RGBA_NV12(int cols,int rows,int src_step,int dst_step,
     }
 }
 
-///////////////////////////////////// RGB <-> YUV //////////////////////////////////////
+///////////////////////////////////// RGB <-> YCrCb //////////////////////////////////////
 
 __constant float c_RGB2YCrCbCoeffs_f[5] = {0.299f, 0.587f, 0.114f, 0.713f, 0.564f};
 __constant int   c_RGB2YCrCbCoeffs_i[5] = {R2Y, G2Y, B2Y, 11682, 9241};
 
-__kernel void RGB2YCrCb(int cols,int rows,int src_step,int dst_step,int channels,
+__kernel void RGB2YCrCb(int cols, int rows, int src_step, int dst_step,
                         int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
                         int src_offset, int dst_offset)
 {
@@ -275,28 +283,735 @@ __kernel void RGB2YCrCb(int cols,int rows,int src_step,int dst_step,int channels
 
     if (y < rows && x < cols)
     {
-        x *= channels;
+        x <<= 2;
         int src_idx = mad24(y, src_step, src_offset + x);
         int dst_idx = mad24(y, dst_step, dst_offset + x);
 
-        dst += dst_idx;
-        const DATA_TYPE rgb[] = { src[src_idx], src[src_idx + 1], src[src_idx + 2] };
+        DATA_TYPE rgb[] = { src[src_idx], src[src_idx + 1], src[src_idx + 2] };
 
-#if defined (DEPTH_5)
+#ifdef DEPTH_5
         __constant float * coeffs = c_RGB2YCrCbCoeffs_f;
-        const DATA_TYPE Y  = rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx];
-        const DATA_TYPE Cr = (rgb[bidx^2] - Y) * coeffs[3] + HALF_MAX;
-        const DATA_TYPE Cb = (rgb[bidx] - Y) * coeffs[4] + HALF_MAX;
+        DATA_TYPE Y  = rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx];
+        DATA_TYPE Cr = (rgb[bidx^2] - Y) * coeffs[3] + HALF_MAX;
+        DATA_TYPE Cb = (rgb[bidx] - Y) * coeffs[4] + HALF_MAX;
 #else
         __constant int * coeffs = c_RGB2YCrCbCoeffs_i;
-        const int delta = HALF_MAX * (1 << yuv_shift);
-        const int Y =  CV_DESCALE(rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx], yuv_shift);
-        const int Cr = CV_DESCALE((rgb[bidx^2] - Y) * coeffs[3] + delta, yuv_shift);
-        const int Cb = CV_DESCALE((rgb[bidx] - Y) * coeffs[4] + delta, yuv_shift);
+        int delta = HALF_MAX * (1 << yuv_shift);
+        int Y =  CV_DESCALE(rgb[0] * coeffs[bidx^2] + rgb[1] * coeffs[1] + rgb[2] * coeffs[bidx], yuv_shift);
+        int Cr = CV_DESCALE((rgb[bidx^2] - Y) * coeffs[3] + delta, yuv_shift);
+        int Cb = CV_DESCALE((rgb[bidx] - Y) * coeffs[4] + delta, yuv_shift);
+#endif
+
+        dst[dst_idx] = SAT_CAST( Y );
+        dst[dst_idx + 1] = SAT_CAST( Cr );
+        dst[dst_idx + 2] = SAT_CAST( Cb );
+    }
+}
+
+__constant float c_YCrCb2RGBCoeffs_f[4] = { 1.403f, -0.714f, -0.344f, 1.773f };
+__constant int   c_YCrCb2RGBCoeffs_i[4] = { 22987, -11698, -5636, 29049 };
+
+__kernel void YCrCb2RGB(int cols, int rows, int src_step, int dst_step,
+                        int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
+                        int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        DATA_TYPE ycrcb[] = { src[src_idx], src[src_idx + 1], src[src_idx + 2] };
+
+#ifdef DEPTH_5
+        __constant float * coeff = c_YCrCb2RGBCoeffs_f;
+        float r = ycrcb[0] + coeff[0] * (ycrcb[1] - HALF_MAX);
+        float g = ycrcb[0] + coeff[1] * (ycrcb[1] - HALF_MAX) + coeff[2] * (ycrcb[2] - HALF_MAX);
+        float b = ycrcb[0] + coeff[3] * (ycrcb[2] - HALF_MAX);
+#else
+        __constant int * coeff = c_YCrCb2RGBCoeffs_i;
+        int r = ycrcb[0] + CV_DESCALE(coeff[0] * (ycrcb[1] - HALF_MAX), yuv_shift);
+        int g = ycrcb[0] + CV_DESCALE(coeff[1] * (ycrcb[1] - HALF_MAX) + coeff[2] * (ycrcb[2] - HALF_MAX), yuv_shift);
+        int b = ycrcb[0] + CV_DESCALE(coeff[3] * (ycrcb[2] - HALF_MAX), yuv_shift);
+#endif
+
+        dst[dst_idx + (bidx^2)] = SAT_CAST(r);
+        dst[dst_idx + 1] = SAT_CAST(g);
+        dst[dst_idx + bidx] = SAT_CAST(b);
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+///////////////////////////////////// RGB <-> XYZ //////////////////////////////////////
+
+__kernel void RGB2XYZ(int cols, int rows, int src_step, int dst_step,
+                      int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
+                      int src_offset, int dst_offset, __constant COEFF_TYPE * coeffs)
+{
+    int dx = get_global_id(0);
+    int dy = get_global_id(1);
+
+    if (dy < rows && dx < cols)
+    {
+        dx <<= 2;
+        int src_idx = mad24(dy, src_step, src_offset + dx);
+        int dst_idx = mad24(dy, dst_step, dst_offset + dx);
+
+        DATA_TYPE r = src[src_idx], g = src[src_idx + 1], b = src[src_idx + 2];
+
+#ifdef DEPTH_5
+        float x = r * coeffs[0] + g * coeffs[1] + b * coeffs[2];
+        float y = r * coeffs[3] + g * coeffs[4] + b * coeffs[5];
+        float z = r * coeffs[6] + g * coeffs[7] + b * coeffs[8];
+#else
+        int x = CV_DESCALE(r * coeffs[0] + g * coeffs[1] + b * coeffs[2], xyz_shift);
+        int y = CV_DESCALE(r * coeffs[3] + g * coeffs[4] + b * coeffs[5], xyz_shift);
+        int z = CV_DESCALE(r * coeffs[6] + g * coeffs[7] + b * coeffs[8], xyz_shift);
+#endif
+        dst[dst_idx] = SAT_CAST(x);
+        dst[dst_idx + 1] = SAT_CAST(y);
+        dst[dst_idx + 2] = SAT_CAST(z);
+    }
+}
+
+__kernel void XYZ2RGB(int cols, int rows, int src_step, int dst_step,
+                      int bidx, __global const DATA_TYPE* src, __global DATA_TYPE* dst,
+                      int src_offset, int dst_offset, __constant COEFF_TYPE * coeffs)
+{
+    int dx = get_global_id(0);
+    int dy = get_global_id(1);
+
+    if (dy < rows && dx < cols)
+    {
+        dx <<= 2;
+        int src_idx = mad24(dy, src_step, src_offset + dx);
+        int dst_idx = mad24(dy, dst_step, dst_offset + dx);
+
+        DATA_TYPE x = src[src_idx], y = src[src_idx + 1], z = src[src_idx + 2];
+
+#ifdef DEPTH_5
+        float b = x * coeffs[0] + y * coeffs[1] + z * coeffs[2];
+        float g = x * coeffs[3] + y * coeffs[4] + z * coeffs[5];
+        float r = x * coeffs[6] + y * coeffs[7] + z * coeffs[8];
+#else
+        int b = CV_DESCALE(x * coeffs[0] + y * coeffs[1] + z * coeffs[2], xyz_shift);
+        int g = CV_DESCALE(x * coeffs[3] + y * coeffs[4] + z * coeffs[5], xyz_shift);
+        int r = CV_DESCALE(x * coeffs[6] + y * coeffs[7] + z * coeffs[8], xyz_shift);
+#endif
+        dst[dst_idx] = SAT_CAST(b);
+        dst[dst_idx + 1] = SAT_CAST(g);
+        dst[dst_idx + 2] = SAT_CAST(r);
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+///////////////////////////////////// RGB[A] <-> BGR[A] //////////////////////////////////////
+
+__kernel void RGB(int cols, int rows, int src_step, int dst_step,
+                  __global const DATA_TYPE * src, __global DATA_TYPE * dst,
+                  int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+#ifdef REVERSE
+        dst[dst_idx] = src[src_idx + 2];
+        dst[dst_idx + 1] = src[src_idx + 1];
+        dst[dst_idx + 2] = src[src_idx];
+#elif defined ORDER
+        dst[dst_idx] = src[src_idx];
+        dst[dst_idx + 1] = src[src_idx + 1];
+        dst[dst_idx + 2] = src[src_idx + 2];
+#endif
+
+#if dcn == 4
+#if scn == 3
+        dst[dst_idx + 3] = MAX_NUM;
+#else
+        dst[dst_idx + 3] = src[src_idx + 3];
+#endif
+#endif
+    }
+}
+
+///////////////////////////////////// RGB5x5 <-> RGB //////////////////////////////////////
+
+__kernel void RGB5x52RGB(int cols, int rows, int src_step, int dst_step, int bidx,
+                         __global const ushort * src, __global uchar * dst,
+                         int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + (x << 2));
+        ushort t = src[src_idx];
+
+#if greenbits == 6
+        dst[dst_idx + bidx] = (uchar)(t << 3);
+        dst[dst_idx + 1] = (uchar)((t >> 3) & ~3);
+        dst[dst_idx + (bidx^2)] = (uchar)((t >> 8) & ~7);
+#else
+        dst[dst_idx + bidx] = (uchar)(t << 3);
+        dst[dst_idx + 1] = (uchar)((t >> 2) & ~7);
+        dst[dst_idx + (bidx^2)] = (uchar)((t >> 7) & ~7);
+#endif
+
+#if dcn == 4
+#if greenbits == 6
+        dst[dst_idx + 3] = 255;
+#else
+        dst[dst_idx + 3] = t & 0x8000 ? 255 : 0;
+#endif
+#endif
+    }
+}
+
+__kernel void RGB2RGB5x5(int cols, int rows, int src_step, int dst_step, int bidx,
+                         __global const uchar * src, __global ushort * dst,
+                         int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        int src_idx = mad24(y, src_step, src_offset + (x << 2));
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+#if greenbits == 6
+            dst[dst_idx] = (ushort)((src[src_idx + bidx] >> 3)|((src[src_idx + 1]&~3) << 3)|((src[src_idx + (bidx^2)]&~7) << 8));
+#elif scn == 3
+            dst[dst_idx] = (ushort)((src[src_idx + bidx] >> 3)|((src[src_idx + 1]&~7) << 2)|((src[src_idx + (bidx^2)]&~7) << 7));
+#else
+            dst[dst_idx] = (ushort)((src[src_idx + bidx] >> 3)|((src[src_idx + 1]&~7) << 2)|
+                ((src[src_idx + (bidx^2)]&~7) << 7)|(src[src_idx + 3] ? 0x8000 : 0));
 #endif
+    }
+}
+
+///////////////////////////////////// RGB5x5 <-> RGB //////////////////////////////////////
+
+__kernel void BGR5x52Gray(int cols, int rows, int src_step, int dst_step, int bidx,
+                          __global const ushort * src, __global uchar * dst,
+                          int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
-        dst[0] = SAT_CAST( Y );
-        dst[1] = SAT_CAST( Cr );
-        dst[2] = SAT_CAST( Cb );
+    if (y < rows && x < cols)
+    {
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+        int t = src[src_idx];
+
+#if greenbits == 6
+        dst[dst_idx] = (uchar)CV_DESCALE(((t << 3) & 0xf8)*B2Y +
+                                         ((t >> 3) & 0xfc)*G2Y +
+                                         ((t >> 8) & 0xf8)*R2Y, yuv_shift);
+#else
+        dst[dst_idx] = (uchar)CV_DESCALE(((t << 3) & 0xf8)*B2Y +
+                                         ((t >> 2) & 0xf8)*G2Y +
+                                         ((t >> 7) & 0xf8)*R2Y, yuv_shift);
+#endif
     }
 }
+
+__kernel void Gray2BGR5x5(int cols, int rows, int src_step, int dst_step, int bidx,
+                          __global const uchar * src, __global ushort * dst,
+                          int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+        int t = src[src_idx];
+
+#if greenbits == 6
+        dst[dst_idx] = (ushort)((t >> 3) | ((t & ~3) << 3) | ((t & ~7) << 8));
+#else
+        t >>= 3;
+        dst[dst_idx] = (ushort)(t|(t << 5)|(t << 10));
+#endif
+    }
+}
+
+///////////////////////////////////// RGB <-> HSV //////////////////////////////////////
+
+__constant int sector_data[][3] = { {1, 3, 0}, { 1, 0, 2 }, { 3, 0, 1 }, { 0, 2, 1 }, { 0, 1, 3 }, { 2, 1, 0 } };
+
+#ifdef DEPTH_0
+
+__kernel void RGB2HSV(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const uchar * src, __global uchar * dst,
+                      int src_offset, int dst_offset,
+                      __constant int * sdiv_table, __constant int * hdiv_table)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        int b = src[src_idx + bidx], g = src[src_idx + 1], r = src[src_idx + (bidx^2)];
+        int h, s, v = b;
+        int vmin = b, diff;
+        int vr, vg;
+
+        v = max( v, g );
+        v = max( v, r );
+        vmin = min( vmin, g );
+        vmin = min( vmin, r );
+
+        diff = v - vmin;
+        vr = v == r ? -1 : 0;
+        vg = v == g ? -1 : 0;
+
+        s = (diff * sdiv_table[v] + (1 << (hsv_shift-1))) >> hsv_shift;
+        h = (vr & (g - b)) +
+            (~vr & ((vg & (b - r + 2 * diff)) + ((~vg) & (r - g + 4 * diff))));
+        h = (h * hdiv_table[diff] + (1 << (hsv_shift-1))) >> hsv_shift;
+        h += h < 0 ? hrange : 0;
+
+        dst[dst_idx] = convert_uchar_sat_rte(h);
+        dst[dst_idx + 1] = (uchar)s;
+        dst[dst_idx + 2] = (uchar)v;
+    }
+}
+
+__kernel void HSV2RGB(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const uchar * src, __global uchar * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float h = src[src_idx], s = src[src_idx + 1]*(1/255.f), v = src[src_idx + 2]*(1/255.f);
+        float b, g, r;
+
+        if (s != 0)
+        {
+            float tab[4];
+            int sector;
+            h *= hscale;
+            if( h < 0 )
+                do h += 6; while( h < 0 );
+            else if( h >= 6 )
+                do h -= 6; while( h >= 6 );
+            sector = convert_int_sat_rtn(h);
+            h -= sector;
+            if( (unsigned)sector >= 6u )
+            {
+                sector = 0;
+                h = 0.f;
+            }
+
+            tab[0] = v;
+            tab[1] = v*(1.f - s);
+            tab[2] = v*(1.f - s*h);
+            tab[3] = v*(1.f - s*(1.f - h));
+
+            b = tab[sector_data[sector][0]];
+            g = tab[sector_data[sector][1]];
+            r = tab[sector_data[sector][2]];
+        }
+        else
+            b = g = r = v;
+
+        dst[dst_idx + bidx] = convert_uchar_sat_rte(b*255.f);
+        dst[dst_idx + 1] = convert_uchar_sat_rte(g*255.f);
+        dst[dst_idx + (bidx^2)] = convert_uchar_sat_rte(r*255.f);
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+#elif defined DEPTH_5
+
+__kernel void RGB2HSV(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const float * src, __global float * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float b = src[src_idx + bidx], g = src[src_idx + 1], r = src[src_idx + (bidx^2)];
+        float h, s, v;
+
+        float vmin, diff;
+
+        v = vmin = r;
+        if( v < g ) v = g;
+        if( v < b ) v = b;
+        if( vmin > g ) vmin = g;
+        if( vmin > b ) vmin = b;
+
+        diff = v - vmin;
+        s = diff/(float)(fabs(v) + FLT_EPSILON);
+        diff = (float)(60./(diff + FLT_EPSILON));
+        if( v == r )
+            h = (g - b)*diff;
+        else if( v == g )
+            h = (b - r)*diff + 120.f;
+        else
+            h = (r - g)*diff + 240.f;
+
+        if( h < 0 ) h += 360.f;
+
+        dst[dst_idx] = h*hscale;
+        dst[dst_idx + 1] = s;
+        dst[dst_idx + 2] = v;
+    }
+}
+
+__kernel void HSV2RGB(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const float * src, __global float * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float h = src[src_idx], s = src[src_idx + 1], v = src[src_idx + 2];
+        float b, g, r;
+
+        if (s != 0)
+        {
+            float tab[4];
+            int sector;
+            h *= hscale;
+            if(h < 0)
+                do h += 6; while (h < 0);
+            else if (h >= 6)
+                do h -= 6; while (h >= 6);
+            sector = convert_int_sat_rtn(h);
+            h -= sector;
+            if ((unsigned)sector >= 6u)
+            {
+                sector = 0;
+                h = 0.f;
+            }
+
+            tab[0] = v;
+            tab[1] = v*(1.f - s);
+            tab[2] = v*(1.f - s*h);
+            tab[3] = v*(1.f - s*(1.f - h));
+
+            b = tab[sector_data[sector][0]];
+            g = tab[sector_data[sector][1]];
+            r = tab[sector_data[sector][2]];
+        }
+        else
+            b = g = r = v;
+
+        dst[dst_idx + bidx] = b;
+        dst[dst_idx + 1] = g;
+        dst[dst_idx + (bidx^2)] = r;
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+#endif
+
+///////////////////////////////////// RGB <-> HLS //////////////////////////////////////
+
+#ifdef DEPTH_0
+
+__kernel void RGB2HLS(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const uchar * src, __global uchar * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float b = src[src_idx + bidx]*(1/255.f), g = src[src_idx + 1]*(1/255.f), r = src[src_idx + (bidx^2)]*(1/255.f);
+        float h = 0.f, s = 0.f, l;
+        float vmin, vmax, diff;
+
+        vmax = vmin = r;
+        if (vmax < g) vmax = g;
+        if (vmax < b) vmax = b;
+        if (vmin > g) vmin = g;
+        if (vmin > b) vmin = b;
+
+        diff = vmax - vmin;
+        l = (vmax + vmin)*0.5f;
+
+        if (diff > FLT_EPSILON)
+        {
+            s = l < 0.5f ? diff/(vmax + vmin) : diff/(2 - vmax - vmin);
+            diff = 60.f/diff;
+
+            if( vmax == r )
+                h = (g - b)*diff;
+            else if( vmax == g )
+                h = (b - r)*diff + 120.f;
+            else
+                h = (r - g)*diff + 240.f;
+
+            if( h < 0.f ) h += 360.f;
+        }
+
+        dst[dst_idx] = convert_uchar_sat_rte(h*hscale);
+        dst[dst_idx + 1] = convert_uchar_sat_rte(l*255.f);
+        dst[dst_idx + 2] = convert_uchar_sat_rte(s*255.f);
+    }
+}
+
+__kernel void HLS2RGB(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const uchar * src, __global uchar * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float h = src[src_idx], l = src[src_idx + 1]*(1.f/255.f), s = src[src_idx + 2]*(1.f/255.f);
+        float b, g, r;
+
+        if (s != 0)
+        {
+            float tab[4];
+
+            float p2 = l <= 0.5f ? l*(1 + s) : l + s - l*s;
+            float p1 = 2*l - p2;
+
+            h *= hscale;
+            if( h < 0 )
+                do h += 6; while( h < 0 );
+            else if( h >= 6 )
+                do h -= 6; while( h >= 6 );
+
+            int sector = convert_int_sat_rtn(h);
+            h -= sector;
+
+            tab[0] = p2;
+            tab[1] = p1;
+            tab[2] = p1 + (p2 - p1)*(1-h);
+            tab[3] = p1 + (p2 - p1)*h;
+
+            b = tab[sector_data[sector][0]];
+            g = tab[sector_data[sector][1]];
+            r = tab[sector_data[sector][2]];
+        }
+        else
+            b = g = r = l;
+
+        dst[dst_idx + bidx] = convert_uchar_sat_rte(b*255.f);
+        dst[dst_idx + 1] = convert_uchar_sat_rte(g*255.f);
+        dst[dst_idx + (bidx^2)] = convert_uchar_sat_rte(r*255.f);
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+#elif defined DEPTH_5
+
+__kernel void RGB2HLS(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const float * src, __global float * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float b = src[src_idx + bidx], g = src[src_idx + 1], r = src[src_idx + (bidx^2)];
+        float h = 0.f, s = 0.f, l;
+        float vmin, vmax, diff;
+
+        vmax = vmin = r;
+        if (vmax < g) vmax = g;
+        if (vmax < b) vmax = b;
+        if (vmin > g) vmin = g;
+        if (vmin > b) vmin = b;
+
+        diff = vmax - vmin;
+        l = (vmax + vmin)*0.5f;
+
+        if (diff > FLT_EPSILON)
+        {
+            s = l < 0.5f ? diff/(vmax + vmin) : diff/(2 - vmax - vmin);
+            diff = 60.f/diff;
+
+            if( vmax == r )
+                h = (g - b)*diff;
+            else if( vmax == g )
+                h = (b - r)*diff + 120.f;
+            else
+                h = (r - g)*diff + 240.f;
+
+            if( h < 0.f ) h += 360.f;
+        }
+
+        dst[dst_idx] = h*hscale;
+        dst[dst_idx + 1] = l;
+        dst[dst_idx + 2] = s;
+    }
+}
+
+__kernel void HLS2RGB(int cols, int rows, int src_step, int dst_step, int bidx,
+                      __global const float * src, __global float * dst,
+                      int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        float h = src[src_idx], l = src[src_idx + 1], s = src[src_idx + 2];
+        float b, g, r;
+
+        if (s != 0)
+        {
+            float tab[4];
+            int sector;
+
+            float p2 = l <= 0.5f ? l*(1 + s) : l + s - l*s;
+            float p1 = 2*l - p2;
+
+            h *= hscale;
+            if( h < 0 )
+                do h += 6; while( h < 0 );
+            else if( h >= 6 )
+                do h -= 6; while( h >= 6 );
+
+            sector = convert_int_sat_rtn(h);
+            h -= sector;
+
+            tab[0] = p2;
+            tab[1] = p1;
+            tab[2] = p1 + (p2 - p1)*(1-h);
+            tab[3] = p1 + (p2 - p1)*h;
+
+            b = tab[sector_data[sector][0]];
+            g = tab[sector_data[sector][1]];
+            r = tab[sector_data[sector][2]];
+        }
+        else
+            b = g = r = l;
+
+        dst[dst_idx + bidx] = b;
+        dst[dst_idx + 1] = g;
+        dst[dst_idx + (bidx^2)] = r;
+#if dcn == 4
+        dst[dst_idx + 3] = MAX_NUM;
+#endif
+    }
+}
+
+#endif
+
+/////////////////////////// RGBA <-> mRGBA (alpha premultiplied) //////////////
+
+#ifdef DEPTH_0
+
+__kernel void RGBA2mRGBA(int cols, int rows, int src_step, int dst_step,
+                        int bidx, __global const uchar * src, __global uchar * dst,
+                        int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        uchar v0 = src[src_idx], v1 = src[src_idx + 1];
+        uchar v2 = src[src_idx + 2], v3 = src[src_idx + 3];
+
+        dst[dst_idx] = (v0 * v3 + HALF_MAX) / MAX_NUM;
+        dst[dst_idx + 1] = (v1 * v3 + HALF_MAX) / MAX_NUM;
+        dst[dst_idx + 2] = (v2 * v3 + HALF_MAX) / MAX_NUM;
+        dst[dst_idx + 3] = v3;
+    }
+}
+
+__kernel void mRGBA2RGBA(int cols, int rows, int src_step, int dst_step, int bidx,
+                        __global const uchar * src, __global uchar * dst,
+                        int src_offset, int dst_offset)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (y < rows && x < cols)
+    {
+        x <<= 2;
+        int src_idx = mad24(y, src_step, src_offset + x);
+        int dst_idx = mad24(y, dst_step, dst_offset + x);
+
+        uchar v0 = src[src_idx], v1 = src[src_idx + 1];
+        uchar v2 = src[src_idx + 2], v3 = src[src_idx + 3];
+        uchar v3_half = v3 / 2;
+
+        dst[dst_idx] = v3 == 0 ? 0 : (v0 * MAX_NUM + v3_half) / v3;
+        dst[dst_idx + 1] = v3 == 0 ? 0 : (v1 * MAX_NUM + v3_half) / v3;
+        dst[dst_idx + 2] = v3 == 0 ? 0 : (v2 * MAX_NUM + v3_half) / v3;
+        dst[dst_idx + 3] = v3;
+    }
+}
+
+#endif

diff --git a/modules/ocl/src/opencl/filtering_boxFilter.cl b/modules/ocl/src/opencl/filtering_boxFilter.cl
index 7f7fd01..96091ce 100644 (file)
--- a/modules/ocl/src/opencl/filtering_boxFilter.cl
+++ b/modules/ocl/src/opencl/filtering_boxFilter.cl
@@ -146,7 +146,11 @@
 #endif
 
 #if USE_DOUBLE
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
 #define FPTYPE double
 #define CONVERT_TO_FPTYPE CAT(convert_double, VEC_SIZE)
 #else

diff --git a/modules/ocl/src/opencl/filtering_filter2D.cl b/modules/ocl/src/opencl/filtering_filter2D.cl
index f966766..fb7dca5 100644 (file)
--- a/modules/ocl/src/opencl/filtering_filter2D.cl
+++ b/modules/ocl/src/opencl/filtering_filter2D.cl
@@ -143,7 +143,11 @@
 #endif
 
 #if USE_DOUBLE
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
 #define FPTYPE double
 #define CONVERT_TO_FPTYPE CAT(convert_double, VEC_SIZE)
 #else

diff --git a/modules/ocl/src/opencl/haarobjectdetect_scaled2.cl b/modules/ocl/src/opencl/haarobjectdetect_scaled2.cl
index 72b9403..a2feb82 100644 (file)
--- a/modules/ocl/src/opencl/haarobjectdetect_scaled2.cl
+++ b/modules/ocl/src/opencl/haarobjectdetect_scaled2.cl
@@ -45,8 +45,6 @@
 //
 //M*/
 
-// Enter your kernel in this window
-//#pragma OPENCL EXTENSION cl_amd_printf:enable
 #define CV_HAAR_FEATURE_MAX           3
 typedef int   sumtype;
 typedef float sqsumtype;
@@ -288,8 +286,8 @@ __kernel void gpuscaleclassifier(global GpuHidHaarTreeNode *orinode, global GpuH
     int counter = get_global_id(0);
     int tr_x[3], tr_y[3], tr_h[3], tr_w[3], i = 0;
     GpuHidHaarTreeNode t1 = *(orinode + counter);
-#pragma unroll
 
+    #pragma unroll
     for (i = 0; i < 3; i++)
     {
         tr_x[i] = (int)(t1.p[i][0] * scale + 0.5f);
@@ -300,8 +298,8 @@ __kernel void gpuscaleclassifier(global GpuHidHaarTreeNode *orinode, global GpuH
 
     t1.weight[0] = -(t1.weight[1] * tr_h[1] * tr_w[1] + t1.weight[2] * tr_h[2] * tr_w[2]) / (tr_h[0] * tr_w[0]);
     counter += nodenum;
-#pragma unroll
 
+    #pragma unroll
     for (i = 0; i < 3; i++)
     {
         newnode[counter].p[i][0] = tr_x[i];

diff --git a/modules/ocl/src/opencl/imgproc_convolve.cl b/modules/ocl/src/opencl/imgproc_convolve.cl
index fb9596e..b8f9742 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_convolve.cl
+++ b/modules/ocl/src/opencl/imgproc_convolve.cl
@@ -43,11 +43,13 @@
 //
 //M*/
 
-#if defined (__ATI__)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
-#elif defined (__NVIDIA__)
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
 /************************************** convolve **************************************/
 

diff --git a/modules/ocl/src/opencl/imgproc_copymakeboder.cl b/modules/ocl/src/opencl/imgproc_copymakeboder.cl
index d97f660..ac149a4 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_copymakeboder.cl
+++ b/modules/ocl/src/opencl/imgproc_copymakeboder.cl
@@ -34,7 +34,7 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)

diff --git a/modules/ocl/src/opencl/imgproc_integral.cl b/modules/ocl/src/opencl/imgproc_integral.cl
index 05e76f9..1d90e50 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_integral.cl
+++ b/modules/ocl/src/opencl/imgproc_integral.cl
@@ -43,13 +43,17 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#define CONVERT(step) ((step)>>1)
+#else
+#define CONVERT(step) ((step))
 #endif
+
 #define LSIZE 256
 #define LSIZE_1 255
 #define LSIZE_2 254
@@ -60,17 +64,17 @@
 #define GET_CONFLICT_OFFSET(lid) ((lid) >> LOG_NUM_BANKS)
 
 
-kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global float *sqsum,
-                          int src_offset,int pre_invalid,int rows,int cols,int src_step,int dst_step)
+kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global TYPE *sqsum,
+                          int src_offset,int pre_invalid,int rows,int cols,int src_step,int dst_step,int dst1_step)
 {
     int lid = get_local_id(0);
     int gid = get_group_id(0);
     int4 src_t[2], sum_t[2];
-    float4 sqsum_t[2];
+    TYPE4 sqsum_t[2];
     __local int4 lm_sum[2][LSIZE + LOG_LSIZE];
-    __local float4 lm_sqsum[2][LSIZE + LOG_LSIZE];
+    __local TYPE4 lm_sqsum[2][LSIZE + LOG_LSIZE];
     __local int* sum_p;
-    __local float* sqsum_p;
+    __local TYPE* sqsum_p;
     src_step = src_step >> 2;
     gid = gid << 1;
     for(int i = 0; i < rows; i =i + LSIZE_1)
@@ -79,17 +83,17 @@ kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global fl
         src_t[1] = (i + lid < rows ? convert_int4(src[src_offset + (lid+i) * src_step + min(gid + 1, cols - 1)]) : 0);
 
         sum_t[0] = (i == 0 ? 0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[0] = (i == 0 ? (float4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[0] = (i == 0 ? (TYPE4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
         sum_t[1] =  (i == 0 ? 0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[1] =  (i == 0 ? (float4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[1] =  (i == 0 ? (TYPE4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
         barrier(CLK_LOCAL_MEM_FENCE);
 
         int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
         lm_sum[0][bf_loc] = src_t[0];
-        lm_sqsum[0][bf_loc] = convert_float4(src_t[0] * src_t[0]);
+        lm_sqsum[0][bf_loc] = convert_TYPE4(src_t[0] * src_t[0]);
 
         lm_sum[1][bf_loc] = src_t[1];
-        lm_sqsum[1][bf_loc] = convert_float4(src_t[1] * src_t[1]);
+        lm_sqsum[1][bf_loc] = convert_TYPE4(src_t[1] * src_t[1]);
 
         int offset = 1;
         for(int d = LSIZE >> 1 ;  d > 0; d>>=1)
@@ -130,7 +134,8 @@ kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global fl
             }
         }
         barrier(CLK_LOCAL_MEM_FENCE);
-        int loc_s0 = gid * dst_step + i + lid - 1 - pre_invalid * dst_step / 4, loc_s1 = loc_s0 + dst_step ;
+        int loc_s0 = gid * dst_step  + i + lid - 1 - pre_invalid * dst_step /4, loc_s1 = loc_s0 + dst_step ;
+        int loc_sq0 = gid * CONVERT(dst1_step) + i + lid - 1 - pre_invalid * dst1_step / sizeof(TYPE),loc_sq1 = loc_sq0 + CONVERT(dst1_step);
         if(lid > 0 && (i+lid) <= rows)
         {
             lm_sum[0][bf_loc] += sum_t[0];
@@ -138,20 +143,20 @@ kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global fl
             lm_sqsum[0][bf_loc] += sqsum_t[0];
             lm_sqsum[1][bf_loc] += sqsum_t[1];
             sum_p = (__local int*)(&(lm_sum[0][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[0][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[0][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 4 + k >= cols + pre_invalid || gid * 4 + k < pre_invalid) continue;
                 sum[loc_s0 + k * dst_step / 4] = sum_p[k];
-                sqsum[loc_s0 + k * dst_step / 4] = sqsum_p[k];
+                sqsum[loc_sq0 + k * dst1_step / sizeof(TYPE)] = sqsum_p[k];
             }
             sum_p = (__local int*)(&(lm_sum[1][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[1][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[1][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 4 + k + 4 >= cols + pre_invalid) break;
                 sum[loc_s1 + k * dst_step / 4] = sum_p[k];
-                sqsum[loc_s1 + k * dst_step / 4] = sqsum_p[k];
+                sqsum[loc_sq1 + k * dst1_step / sizeof(TYPE)] = sqsum_p[k];
             }
         }
         barrier(CLK_LOCAL_MEM_FENCE);
@@ -159,30 +164,32 @@ kernel void integral_cols_D4(__global uchar4 *src,__global int *sum ,__global fl
 }
 
 
-kernel void integral_rows_D4(__global int4 *srcsum,__global float4 * srcsqsum,__global int *sum ,
-                          __global float *sqsum,int rows,int cols,int src_step,int sum_step,
+kernel void integral_rows_D4(__global int4 *srcsum,__global TYPE4 * srcsqsum,__global int *sum ,
+                          __global TYPE *sqsum,int rows,int cols,int src_step,int src1_step,int sum_step,
                           int sqsum_step,int sum_offset,int sqsum_offset)
 {
     int lid = get_local_id(0);
     int gid = get_group_id(0);
     int4 src_t[2], sum_t[2];
-    float4 sqsrc_t[2],sqsum_t[2];
+    TYPE4 sqsrc_t[2],sqsum_t[2];
     __local int4 lm_sum[2][LSIZE + LOG_LSIZE];
-    __local float4 lm_sqsum[2][LSIZE + LOG_LSIZE];
+    __local TYPE4 lm_sqsum[2][LSIZE + LOG_LSIZE];
     __local int *sum_p;
-    __local float *sqsum_p;
+    __local TYPE *sqsum_p;
     src_step = src_step >> 4;
+    src1_step = (src1_step / sizeof(TYPE)) >> 2 ;
+    gid <<= 1;
     for(int i = 0; i < rows; i =i + LSIZE_1)
     {
-        src_t[0] = i + lid < rows ? srcsum[(lid+i) * src_step + gid * 2] : (int4)0;
-        sqsrc_t[0] = i + lid < rows ? srcsqsum[(lid+i) * src_step + gid * 2] : (float4)0;
-        src_t[1] = i + lid < rows ? srcsum[(lid+i) * src_step + gid * 2 + 1] : (int4)0;
-        sqsrc_t[1] = i + lid < rows ? srcsqsum[(lid+i) * src_step + gid * 2 + 1] : (float4)0;
+        src_t[0] = i + lid < rows ? srcsum[(lid+i) * src_step + gid ] : (int4)0;
+        sqsrc_t[0] = i + lid < rows ? srcsqsum[(lid+i) * src1_step + gid ] : (TYPE4)0;
+        src_t[1] = i + lid < rows ? srcsum[(lid+i) * src_step + gid  + 1] : (int4)0;
+        sqsrc_t[1] = i + lid < rows ? srcsqsum[(lid+i) * src1_step + gid  + 1] : (TYPE4)0;
 
         sum_t[0] =  (i == 0 ? 0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[0] =  (i == 0 ? (float4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[0] =  (i == 0 ? (TYPE4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
         sum_t[1] =  (i == 0 ? 0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[1] =  (i == 0 ? (float4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[1] =  (i == 0 ? (TYPE4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
         barrier(CLK_LOCAL_MEM_FENCE);
 
         int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
@@ -238,17 +245,18 @@ kernel void integral_rows_D4(__global int4 *srcsum,__global float4 * srcsqsum,__
         }
         if(i + lid == 0)
         {
-            int loc0 = gid * 2 * sum_step;
-            int loc1 = gid * 2 * sqsum_step;
+            int loc0 = gid  * sum_step;
+            int loc1 = gid  * CONVERT(sqsum_step);
             for(int k = 1; k <= 8; k++)
             {
-                if(gid * 8 + k > cols) break;
+                if(gid * 4 + k > cols) break;
                 sum[sum_offset + loc0 + k * sum_step / 4] = 0;
-                sqsum[sqsum_offset + loc1 + k * sqsum_step / 4] = 0;
+                sqsum[sqsum_offset + loc1 + k * sqsum_step / sizeof(TYPE)] = 0;
             }
         }
-        int loc_s0 = sum_offset + gid * 2 * sum_step + sum_step / 4 + i + lid, loc_s1 = loc_s0 + sum_step ;
-        int loc_sq0 = sqsum_offset + gid * 2 * sqsum_step + sqsum_step / 4 + i + lid, loc_sq1 = loc_sq0 + sqsum_step ;
+        int loc_s0 = sum_offset + gid  * sum_step + sum_step / 4 + i + lid, loc_s1 = loc_s0 + sum_step ;
+        int loc_sq0 = sqsum_offset + gid  * CONVERT(sqsum_step) + sqsum_step / sizeof(TYPE) + i + lid, loc_sq1 = loc_sq0 + CONVERT(sqsum_step) ;
+
         if(lid > 0 && (i+lid) <= rows)
         {
             lm_sum[0][bf_loc] += sum_t[0];
@@ -256,37 +264,37 @@ kernel void integral_rows_D4(__global int4 *srcsum,__global float4 * srcsqsum,__
             lm_sqsum[0][bf_loc] += sqsum_t[0];
             lm_sqsum[1][bf_loc] += sqsum_t[1];
             sum_p = (__local int*)(&(lm_sum[0][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[0][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[0][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
-                if(gid * 8 + k >= cols) break;
+                if(gid * 4 + k >= cols) break;
                 sum[loc_s0 + k * sum_step / 4] = sum_p[k];
-                sqsum[loc_sq0 + k * sqsum_step / 4] = sqsum_p[k];
+                sqsum[loc_sq0 + k * sqsum_step / sizeof(TYPE)] = sqsum_p[k];
             }
             sum_p = (__local int*)(&(lm_sum[1][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[1][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[1][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
-                if(gid * 8 + 4 + k >= cols) break;
+                if(gid * 4 + 4 + k >= cols) break;
                 sum[loc_s1 + k * sum_step / 4] = sum_p[k];
-                sqsum[loc_sq1 + k * sqsum_step / 4] = sqsum_p[k];
+                sqsum[loc_sq1 + k * sqsum_step / sizeof(TYPE)] = sqsum_p[k];
             }
-        }
+          }
         barrier(CLK_LOCAL_MEM_FENCE);
     }
 }
 
-kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global float *sqsum,
-                          int src_offset,int pre_invalid,int rows,int cols,int src_step,int dst_step)
+kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global TYPE *sqsum,
+                          int src_offset,int pre_invalid,int rows,int cols,int src_step,int dst_step, int dst1_step)
 {
     int lid = get_local_id(0);
     int gid = get_group_id(0);
     float4 src_t[2], sum_t[2];
-    float4 sqsum_t[2];
+    TYPE4 sqsum_t[2];
     __local float4 lm_sum[2][LSIZE + LOG_LSIZE];
-    __local float4 lm_sqsum[2][LSIZE + LOG_LSIZE];
+    __local TYPE4 lm_sqsum[2][LSIZE + LOG_LSIZE];
     __local float* sum_p;
-    __local float* sqsum_p;
+    __local TYPE* sqsum_p;
     src_step = src_step >> 2;
     gid = gid << 1;
     for(int i = 0; i < rows; i =i + LSIZE_1)
@@ -295,17 +303,17 @@ kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global
         src_t[1] = (i + lid < rows ? convert_float4(src[src_offset + (lid+i) * src_step + min(gid + 1, cols - 1)]) : (float4)0);
 
         sum_t[0] = (i == 0 ? (float4)0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[0] = (i == 0 ? (float4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[0] = (i == 0 ? (TYPE4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
         sum_t[1] =  (i == 0 ? (float4)0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[1] =  (i == 0 ? (float4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[1] =  (i == 0 ? (TYPE4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
         barrier(CLK_LOCAL_MEM_FENCE);
 
         int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
         lm_sum[0][bf_loc] = src_t[0];
-        lm_sqsum[0][bf_loc] = convert_float4(src_t[0] * src_t[0]);
+        lm_sqsum[0][bf_loc] = convert_TYPE4(src_t[0] * src_t[0]);
 
         lm_sum[1][bf_loc] = src_t[1];
-        lm_sqsum[1][bf_loc] = convert_float4(src_t[1] * src_t[1]);
+        lm_sqsum[1][bf_loc] = convert_TYPE4(src_t[1] * src_t[1]);
 
         int offset = 1;
         for(int d = LSIZE >> 1 ;  d > 0; d>>=1)
@@ -347,6 +355,7 @@ kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global
         }
         barrier(CLK_LOCAL_MEM_FENCE);
         int loc_s0 = gid * dst_step + i + lid - 1 - pre_invalid * dst_step / 4, loc_s1 = loc_s0 + dst_step ;
+        int loc_sq0 = gid * CONVERT(dst1_step) + i + lid - 1 - pre_invalid * dst1_step / sizeof(TYPE), loc_sq1 = loc_sq0 + CONVERT(dst1_step);
         if(lid > 0 && (i+lid) <= rows)
         {
             lm_sum[0][bf_loc] += sum_t[0];
@@ -354,20 +363,20 @@ kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global
             lm_sqsum[0][bf_loc] += sqsum_t[0];
             lm_sqsum[1][bf_loc] += sqsum_t[1];
             sum_p = (__local float*)(&(lm_sum[0][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[0][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[0][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 4 + k >= cols + pre_invalid || gid * 4 + k < pre_invalid) continue;
                 sum[loc_s0 + k * dst_step / 4] = sum_p[k];
-                sqsum[loc_s0 + k * dst_step / 4] = sqsum_p[k];
+                sqsum[loc_sq0 + k * dst1_step / sizeof(TYPE)] = sqsum_p[k];
             }
             sum_p = (__local float*)(&(lm_sum[1][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[1][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[1][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 4 + k + 4 >= cols + pre_invalid) break;
                 sum[loc_s1 + k * dst_step / 4] = sum_p[k];
-                sqsum[loc_s1 + k * dst_step / 4] = sqsum_p[k];
+                sqsum[loc_sq1 + k * dst1_step / sizeof(TYPE)] = sqsum_p[k];
             }
         }
         barrier(CLK_LOCAL_MEM_FENCE);
@@ -375,30 +384,31 @@ kernel void integral_cols_D5(__global uchar4 *src,__global float *sum ,__global
 }
 
 
-kernel void integral_rows_D5(__global float4 *srcsum,__global float4 * srcsqsum,__global float *sum ,
-                          __global float *sqsum,int rows,int cols,int src_step,int sum_step,
+kernel void integral_rows_D5(__global float4 *srcsum,__global TYPE4 * srcsqsum,__global float *sum ,
+                          __global TYPE *sqsum,int rows,int cols,int src_step,int src1_step, int sum_step,
                           int sqsum_step,int sum_offset,int sqsum_offset)
 {
     int lid = get_local_id(0);
     int gid = get_group_id(0);
     float4 src_t[2], sum_t[2];
-    float4 sqsrc_t[2],sqsum_t[2];
+    TYPE4 sqsrc_t[2],sqsum_t[2];
     __local float4 lm_sum[2][LSIZE + LOG_LSIZE];
-    __local float4 lm_sqsum[2][LSIZE + LOG_LSIZE];
+    __local TYPE4 lm_sqsum[2][LSIZE + LOG_LSIZE];
     __local float *sum_p;
-    __local float *sqsum_p;
+    __local TYPE *sqsum_p;
     src_step = src_step >> 4;
+    src1_step = (src1_step / sizeof(TYPE)) >> 2;
     for(int i = 0; i < rows; i =i + LSIZE_1)
     {
         src_t[0] = i + lid < rows ? srcsum[(lid+i) * src_step + gid * 2] : (float4)0;
-        sqsrc_t[0] = i + lid < rows ? srcsqsum[(lid+i) * src_step + gid * 2] : (float4)0;
+        sqsrc_t[0] = i + lid < rows ? srcsqsum[(lid+i) * src1_step + gid * 2] : (TYPE4)0;
         src_t[1] = i + lid < rows ? srcsum[(lid+i) * src_step + gid * 2 + 1] : (float4)0;
-        sqsrc_t[1] = i + lid < rows ? srcsqsum[(lid+i) * src_step + gid * 2 + 1] : (float4)0;
+        sqsrc_t[1] = i + lid < rows ? srcsqsum[(lid+i) * src1_step + gid * 2 + 1] : (TYPE4)0;
 
         sum_t[0] =  (i == 0 ? (float4)0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[0] =  (i == 0 ? (float4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[0] =  (i == 0 ? (TYPE4)0 : lm_sqsum[0][LSIZE_2 + LOG_LSIZE]);
         sum_t[1] =  (i == 0 ? (float4)0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
-        sqsum_t[1] =  (i == 0 ? (float4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
+        sqsum_t[1] =  (i == 0 ? (TYPE4)0 : lm_sqsum[1][LSIZE_2 + LOG_LSIZE]);
         barrier(CLK_LOCAL_MEM_FENCE);
 
         int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
@@ -455,16 +465,16 @@ kernel void integral_rows_D5(__global float4 *srcsum,__global float4 * srcsqsum,
         if(i + lid == 0)
         {
             int loc0 = gid * 2 * sum_step;
-            int loc1 = gid * 2 * sqsum_step;
+            int loc1 = gid * 2 * CONVERT(sqsum_step);
             for(int k = 1; k <= 8; k++)
             {
                 if(gid * 8 + k > cols) break;
                 sum[sum_offset + loc0 + k * sum_step / 4] = 0;
-                sqsum[sqsum_offset + loc1 + k * sqsum_step / 4] = 0;
+                sqsum[sqsum_offset + loc1 + k * sqsum_step / sizeof(TYPE)] = 0;
             }
         }
         int loc_s0 = sum_offset + gid * 2 * sum_step + sum_step / 4 + i + lid, loc_s1 = loc_s0 + sum_step ;
-        int loc_sq0 = sqsum_offset + gid * 2 * sqsum_step + sqsum_step / 4 + i + lid, loc_sq1 = loc_sq0 + sqsum_step ;
+        int loc_sq0 = sqsum_offset + gid * 2 * CONVERT(sqsum_step) + sqsum_step / sizeof(TYPE) + i + lid, loc_sq1 = loc_sq0 + CONVERT(sqsum_step) ;
         if(lid > 0 && (i+lid) <= rows)
         {
             lm_sum[0][bf_loc] += sum_t[0];
@@ -472,20 +482,20 @@ kernel void integral_rows_D5(__global float4 *srcsum,__global float4 * srcsqsum,
             lm_sqsum[0][bf_loc] += sqsum_t[0];
             lm_sqsum[1][bf_loc] += sqsum_t[1];
             sum_p = (__local float*)(&(lm_sum[0][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[0][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[0][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 8 + k >= cols) break;
                 sum[loc_s0 + k * sum_step / 4] = sum_p[k];
-                sqsum[loc_sq0 + k * sqsum_step / 4] = sqsum_p[k];
+                sqsum[loc_sq0 + k * sqsum_step / sizeof(TYPE)] = sqsum_p[k];
             }
             sum_p = (__local float*)(&(lm_sum[1][bf_loc]));
-            sqsum_p = (__local float*)(&(lm_sqsum[1][bf_loc]));
+            sqsum_p = (__local TYPE*)(&(lm_sqsum[1][bf_loc]));
             for(int k = 0; k < 4; k++)
             {
                 if(gid * 8 + 4 + k >= cols) break;
                 sum[loc_s1 + k * sum_step / 4] = sum_p[k];
-                sqsum[loc_sq1 + k * sqsum_step / 4] = sqsum_p[k];
+                sqsum[loc_sq1 + k * sqsum_step / sizeof(TYPE)] = sqsum_p[k];
             }
         }
         barrier(CLK_LOCAL_MEM_FENCE);

diff --git a/modules/ocl/src/opencl/imgproc_integral_sum.cl b/modules/ocl/src/opencl/imgproc_integral_sum.cl
index a6f73c7..6624061 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_integral_sum.cl
+++ b/modules/ocl/src/opencl/imgproc_integral_sum.cl
@@ -43,11 +43,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/imgproc_remap.cl b/modules/ocl/src/opencl/imgproc_remap.cl
index 53c0539..e1e3ca8 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_remap.cl
+++ b/modules/ocl/src/opencl/imgproc_remap.cl
@@ -43,14 +43,21 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 
+enum
+{
+    INTER_BITS = 5,
+    INTER_TAB_SIZE = 1 << INTER_BITS,
+    INTER_TAB_SIZE2 = INTER_TAB_SIZE * INTER_TAB_SIZE
+};
+
 #ifdef INTER_NEAREST
 #define convertToWT
 #endif
@@ -204,8 +211,92 @@ __kernel void remap_16SC2(__global const T * restrict src, __global T * dst, __g
     }
 }
 
+__kernel void remap_16SC2_16UC1(__global const T * restrict src, __global T * dst, __global short2 * map1, __global ushort * map2,
+        int src_offset, int dst_offset, int map1_offset, int map2_offset,
+        int src_step, int dst_step, int map1_step, int map2_step,
+        int src_cols, int src_rows, int dst_cols, int dst_rows, T scalar)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (x < dst_cols && y < dst_rows)
+    {
+        int dstIdx = mad24(y, dst_step, x + dst_offset);
+        int map1Idx = mad24(y, map1_step, x + map1_offset);
+        int map2Idx = mad24(y, map2_step, x + map2_offset);
+
+        int map2Value = convert_int(map2[map2Idx]) & (INTER_TAB_SIZE2 - 1);
+        int dx = (map2Value & (INTER_TAB_SIZE - 1)) < (INTER_TAB_SIZE >> 1) ? 1 : 0;
+        int dy = (map2Value >> INTER_BITS) < (INTER_TAB_SIZE >> 1) ? 1 : 0;
+        int2 gxy = convert_int2(map1[map1Idx]) + (int2)(dx, dy);
+        int gx = gxy.x, gy = gxy.y;
+
+        if (NEED_EXTRAPOLATION(gx, gy))
+            EXTRAPOLATE(gxy, dst[dstIdx])
+        else
+        {
+            int srcIdx = mad24(gy, src_step, gx + src_offset);
+            dst[dstIdx] = src[srcIdx];
+        }
+    }
+}
+
 #elif INTER_LINEAR
 
+__kernel void remap_16SC2_16UC1(__global T const * restrict src, __global T * dst,
+        __global short2 * restrict map1, __global ushort * restrict map2,
+        int src_offset, int dst_offset, int map1_offset, int map2_offset,
+        int src_step, int dst_step, int map1_step, int map2_step,
+        int src_cols, int src_rows, int dst_cols, int dst_rows, T nVal)
+{
+    int x = get_global_id(0);
+    int y = get_global_id(1);
+
+    if (x < dst_cols && y < dst_rows)
+    {
+        int dstIdx = mad24(y, dst_step, x + dst_offset);
+        int map1Idx = mad24(y, map1_step, x + map1_offset);
+        int map2Idx = mad24(y, map2_step, x + map2_offset);
+
+        int2 map_dataA = convert_int2(map1[map1Idx]);
+        int2 map_dataB = (int2)(map_dataA.x + 1, map_dataA.y);
+        int2 map_dataC = (int2)(map_dataA.x, map_dataA.y + 1);
+        int2 map_dataD = (int2)(map_dataA.x + 1, map_dataA.y + 1);
+
+        ushort map2Value = (ushort)(map2[map2Idx] & (INTER_TAB_SIZE2 - 1));
+        WT2 u = (WT2)(map2Value & (INTER_TAB_SIZE - 1), map2Value >> INTER_BITS) / (WT2)(INTER_TAB_SIZE);
+
+        WT scalar = convertToWT(nVal);
+        WT a = scalar, b = scalar, c = scalar, d = scalar;
+
+        if (!NEED_EXTRAPOLATION(map_dataA.x, map_dataA.y))
+            a = convertToWT(src[mad24(map_dataA.y, src_step, map_dataA.x + src_offset)]);
+        else
+            EXTRAPOLATE(map_dataA, a);
+
+        if (!NEED_EXTRAPOLATION(map_dataB.x, map_dataB.y))
+            b = convertToWT(src[mad24(map_dataB.y, src_step, map_dataB.x + src_offset)]);
+        else
+            EXTRAPOLATE(map_dataB, b);
+
+        if (!NEED_EXTRAPOLATION(map_dataC.x, map_dataC.y))
+            c = convertToWT(src[mad24(map_dataC.y, src_step, map_dataC.x + src_offset)]);
+        else
+            EXTRAPOLATE(map_dataC, c);
+
+        if (!NEED_EXTRAPOLATION(map_dataD.x, map_dataD.y))
+            d = convertToWT(src[mad24(map_dataD.y, src_step, map_dataD.x + src_offset)]);
+        else
+            EXTRAPOLATE(map_dataD, d);
+
+        WT dst_data = a * (1 - u.x) * (1 - u.y) +
+                      b * (u.x)     * (1 - u.y) +
+                      c * (1 - u.x) * (u.y) +
+                      d * (u.x)     * (u.y);
+        dst[dstIdx] = convertToT(dst_data);
+    }
+}
+
 __kernel void remap_2_32FC1(__global T const * restrict  src, __global T * dst,
         __global float * map1, __global float * map2,
         int src_offset, int dst_offset, int map1_offset, int map2_offset,
@@ -226,10 +317,10 @@ __kernel void remap_2_32FC1(__global T const * restrict  src, __global T * dst,
         int2 map_dataA = convert_int2_sat_rtn(map_data);
         int2 map_dataB = (int2)(map_dataA.x + 1, map_dataA.y);
         int2 map_dataC = (int2)(map_dataA.x, map_dataA.y + 1);
-        int2 map_dataD = (int2)(map_dataA.x + 1, map_dataA.y +1);
+        int2 map_dataD = (int2)(map_dataA.x + 1, map_dataA.y + 1);
 
         float2 _u = map_data - convert_float2(map_dataA);
-        WT2 u = convertToWT2(convert_int2_rte(convertToWT2(_u) * (WT2)32)) / (WT2)32;
+        WT2 u = convertToWT2(convert_int2_rte(convertToWT2(_u) * (WT2)INTER_TAB_SIZE)) / (WT2)INTER_TAB_SIZE;
         WT scalar = convertToWT(nVal);
         WT a = scalar, b = scalar, c = scalar, d = scalar;
 
@@ -253,10 +344,10 @@ __kernel void remap_2_32FC1(__global T const * restrict  src, __global T * dst,
         else
             EXTRAPOLATE(map_dataD, d);
 
-        WT dst_data = a * (WT)(1 - u.x) * (WT)(1 - u.y) +
-                      b * (WT)(u.x)     * (WT)(1 - u.y) +
-                      c * (WT)(1 - u.x) * (WT)(u.y) +
-                      d * (WT)(u.x)     * (WT)(u.y);
+        WT dst_data = a * (1 - u.x) * (1 - u.y) +
+                      b * (u.x)     * (1 - u.y) +
+                      c * (1 - u.x) * (u.y) +
+                      d * (u.x)     * (u.y);
         dst[dstIdx] = convertToT(dst_data);
     }
 }
@@ -282,7 +373,7 @@ __kernel void remap_32FC2(__global T const * restrict  src, __global T * dst,
         int2 map_dataD = (int2)(map_dataA.x + 1, map_dataA.y + 1);
 
         float2 _u = map_data - convert_float2(map_dataA);
-        WT2 u = convertToWT2(convert_int2_rte(convertToWT2(_u) * (WT2)32)) / (WT2)32;
+        WT2 u = convertToWT2(convert_int2_rte(convertToWT2(_u) * (WT2)INTER_TAB_SIZE)) / (WT2)INTER_TAB_SIZE;
         WT scalar = convertToWT(nVal);
         WT a = scalar, b = scalar, c = scalar, d = scalar;
 
@@ -306,10 +397,10 @@ __kernel void remap_32FC2(__global T const * restrict  src, __global T * dst,
         else
             EXTRAPOLATE(map_dataD, d);
 
-        WT dst_data = a * (WT)(1 - u.x) * (WT)(1 - u.y) +
-                      b * (WT)(u.x)     * (WT)(1 - u.y) +
-                      c * (WT)(1 - u.x) * (WT)(u.y) +
-                      d * (WT)(u.x)     * (WT)(u.y);
+        WT dst_data = a * (1 - u.x) * (1 - u.y) +
+                      b * (u.x)     * (1 - u.y) +
+                      c * (1 - u.x) * (u.y) +
+                      d * (u.x)     * (u.y);
         dst[dstIdx] = convertToT(dst_data);
     }
 }

diff --git a/modules/ocl/src/opencl/imgproc_resize.cl b/modules/ocl/src/opencl/imgproc_resize.cl
index 2bb75b9..ebf8c71 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_resize.cl
+++ b/modules/ocl/src/opencl/imgproc_resize.cl
@@ -45,16 +45,16 @@
 
 
 // resize kernel
-// Currently, CV_8UC1  CV_8UC4  CV_32FC1 and CV_32FC4are supported.
+// Currently, CV_8UC1, CV_8UC4, CV_32FC1 and CV_32FC4 are supported.
 // We shall support other types later if necessary.
 
-#if defined DOUBLE_SUPPORT
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
-#define F double
-#else
-#define F float
 #endif
-
+#endif
 
 #define INTER_RESIZE_COEF_BITS 11
 #define INTER_RESIZE_COEF_SCALE (1 << INTER_RESIZE_COEF_BITS)
@@ -62,8 +62,10 @@
 #define CAST_SCALE (1.0f/(1<<CAST_BITS))
 #define INC(x,l) ((x+1) >= (l) ? (x):((x)+1))
 
+#ifdef LN
+
 __kernel void resizeLN_C1_D0(__global uchar * dst, __global uchar const * restrict src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
+                     int dst_offset, int src_offset,int dst_step, int src_step,
                      int src_cols, int src_rows, int dst_cols, int dst_rows, float ifx, float ify )
 {
     int gx = get_global_id(0);
@@ -71,7 +73,7 @@ __kernel void resizeLN_C1_D0(__global uchar * dst, __global uchar const * restri
 
     float4  sx, u, xf;
     int4 x, DX;
-    gx = (gx<<2) - (dstoffset_in_pixel&3);
+    gx = (gx<<2) - (dst_offset&3);
     DX = (int4)(gx, gx+1, gx+2, gx+3);
     sx = (convert_float4(DX) + 0.5f) * ifx - 0.5f;
     xf = floor(sx);
@@ -109,10 +111,10 @@ __kernel void resizeLN_C1_D0(__global uchar * dst, __global uchar const * restri
     int4 val1, val2, val;
     int4 sdata1, sdata2, sdata3, sdata4;
 
-    int4 pos1 = mad24((int4)y, (int4)srcstep_in_pixel, x+(int4)srcoffset_in_pixel);
-    int4 pos2 = mad24((int4)y, (int4)srcstep_in_pixel, x_+(int4)srcoffset_in_pixel);
-    int4 pos3 = mad24((int4)y_, (int4)srcstep_in_pixel, x+(int4)srcoffset_in_pixel);
-    int4 pos4 = mad24((int4)y_, (int4)srcstep_in_pixel, x_+(int4)srcoffset_in_pixel);
+    int4 pos1 = mad24((int4)y, (int4)src_step, x+(int4)src_offset);
+    int4 pos2 = mad24((int4)y, (int4)src_step, x_+(int4)src_offset);
+    int4 pos3 = mad24((int4)y_, (int4)src_step, x+(int4)src_offset);
+    int4 pos4 = mad24((int4)y_, (int4)src_step, x_+(int4)src_offset);
 
     sdata1.s0 = src[pos1.s0];
     sdata1.s1 = src[pos1.s1];
@@ -140,12 +142,12 @@ __kernel void resizeLN_C1_D0(__global uchar * dst, __global uchar const * restri
 
     val = ((val + (1<<(CAST_BITS-1))) >> CAST_BITS);
 
-    pos4 = mad24(dy, dststep_in_pixel, gx+dstoffset_in_pixel);
+    pos4 = mad24(dy, dst_step, gx+dst_offset);
     pos4.y++;
     pos4.z+=2;
     pos4.w+=3;
     uchar4 uval = convert_uchar4_sat(val);
-        int con = (gx >= 0 && gx+3 < dst_cols && dy >= 0 && dy < dst_rows && (dstoffset_in_pixel&3)==0);
+        int con = (gx >= 0 && gx+3 < dst_cols && dy >= 0 && dy < dst_rows && (dst_offset&3)==0);
     if(con)
     {
         *(__global uchar4*)(dst + pos4.x)=uval;
@@ -172,7 +174,7 @@ __kernel void resizeLN_C1_D0(__global uchar * dst, __global uchar const * restri
 }
 
 __kernel void resizeLN_C4_D0(__global uchar4 * dst, __global uchar4 * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
+                     int dst_offset, int src_offset,int dst_step, int src_step,
                      int src_cols, int src_rows, int dst_cols, int dst_rows, float ifx, float ify )
 {
     int dx = get_global_id(0);
@@ -198,24 +200,24 @@ __kernel void resizeLN_C4_D0(__global uchar4 * dst, __global uchar4 * src,
     int y_ = INC(y,src_rows);
     int x_ = INC(x,src_cols);
     int4 srcpos;
-    srcpos.x = mad24(y, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.y = mad24(y, srcstep_in_pixel, x_+srcoffset_in_pixel);
-    srcpos.z = mad24(y_, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.w = mad24(y_, srcstep_in_pixel, x_+srcoffset_in_pixel);
+    srcpos.x = mad24(y, src_step, x+src_offset);
+    srcpos.y = mad24(y, src_step, x_+src_offset);
+    srcpos.z = mad24(y_, src_step, x+src_offset);
+    srcpos.w = mad24(y_, src_step, x_+src_offset);
     int4 data0 = convert_int4(src[srcpos.x]);
     int4 data1 = convert_int4(src[srcpos.y]);
     int4 data2 = convert_int4(src[srcpos.z]);
     int4 data3 = convert_int4(src[srcpos.w]);
     int4 val = mul24((int4)mul24(U1, V1) ,  data0) + mul24((int4)mul24(U, V1) ,  data1)
                +mul24((int4)mul24(U1, V) ,  data2)+mul24((int4)mul24(U, V) ,  data3);
-    int dstpos = mad24(dy, dststep_in_pixel, dx+dstoffset_in_pixel);
+    int dstpos = mad24(dy, dst_step, dx+dst_offset);
     uchar4 uval =   convert_uchar4((val + (1<<(CAST_BITS-1)))>>CAST_BITS);
     if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
          dst[dstpos] = uval;
 }
 
 __kernel void resizeLN_C1_D5(__global float * dst, __global float * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
+                     int dst_offset, int src_offset,int dst_step, int src_step,
                      int src_cols, int src_rows, int dst_cols, int dst_rows, float ifx, float ify )
 {
     int dx = get_global_id(0);
@@ -235,10 +237,10 @@ __kernel void resizeLN_C1_D5(__global float * dst, __global float * src,
     float u1 = 1.f-u;
     float v1 = 1.f-v;
     int4 srcpos;
-    srcpos.x = mad24(y, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.y = mad24(y, srcstep_in_pixel, x_+srcoffset_in_pixel);
-    srcpos.z = mad24(y_, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.w = mad24(y_, srcstep_in_pixel, x_+srcoffset_in_pixel);
+    srcpos.x = mad24(y, src_step, x+src_offset);
+    srcpos.y = mad24(y, src_step, x_+src_offset);
+    srcpos.z = mad24(y_, src_step, x+src_offset);
+    srcpos.w = mad24(y_, src_step, x_+src_offset);
     float data0 = src[srcpos.x];
     float data1 = src[srcpos.y];
     float data2 = src[srcpos.z];
@@ -248,13 +250,13 @@ __kernel void resizeLN_C1_D5(__global float * dst, __global float * src,
     float val2 = u1 *  data2 +
                 u *  data3;
     float val = v1 * val1 + v * val2;
-    int dstpos = mad24(dy, dststep_in_pixel, dx+dstoffset_in_pixel);
+    int dstpos = mad24(dy, dst_step, dx+dst_offset);
     if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
          dst[dstpos] = val;
 }
 
 __kernel void resizeLN_C4_D5(__global float4 * dst, __global float4 * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
+                     int dst_offset, int src_offset,int dst_step, int src_step,
                      int src_cols, int src_rows, int dst_cols, int dst_rows, float ifx, float ify )
 {
     int dx = get_global_id(0);
@@ -274,10 +276,10 @@ __kernel void resizeLN_C4_D5(__global float4 * dst, __global float4 * src,
     float u1 = 1.f-u;
     float v1 = 1.f-v;
     int4 srcpos;
-    srcpos.x = mad24(y, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.y = mad24(y, srcstep_in_pixel, x_+srcoffset_in_pixel);
-    srcpos.z = mad24(y_, srcstep_in_pixel, x+srcoffset_in_pixel);
-    srcpos.w = mad24(y_, srcstep_in_pixel, x_+srcoffset_in_pixel);
+    srcpos.x = mad24(y, src_step, x+src_offset);
+    srcpos.y = mad24(y, src_step, x_+src_offset);
+    srcpos.z = mad24(y_, src_step, x+src_offset);
+    srcpos.w = mad24(y_, src_step, x_+src_offset);
     float4 s_data1, s_data2, s_data3, s_data4;
     s_data1 = src[srcpos.x];
     s_data2 = src[srcpos.y];
@@ -285,129 +287,119 @@ __kernel void resizeLN_C4_D5(__global float4 * dst, __global float4 * src,
     s_data4 = src[srcpos.w];
     float4 val = u1 * v1 * s_data1 + u * v1 * s_data2
               +u1 * v *s_data3 + u * v *s_data4;
-    int dstpos = mad24(dy, dststep_in_pixel, dx+dstoffset_in_pixel);
+    int dstpos = mad24(dy, dst_step, dx+dst_offset);
 
     if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
          dst[dstpos] = val;
 }
 
-__kernel void resizeNN_C1_D0(__global uchar * dst, __global uchar * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
-                     int src_cols, int src_rows, int dst_cols, int dst_rows, F ifx, F ify )
+#elif defined NN
+
+__kernel void resizeNN(__global T * dst, __global T * src,
+                       int dst_offset, int src_offset, int dst_step, int src_step,
+                       int src_cols, int src_rows, int dst_cols, int dst_rows, float ifx, float ify)
 {
-    int gx = get_global_id(0);
+    int dx = get_global_id(0);
     int dy = get_global_id(1);
 
-    gx = (gx<<2) - (dstoffset_in_pixel&3);
-    //int4 GX = (int4)(gx, gx+1, gx+2, gx+3);
-
-    int4 sx;
-    int sy;
-    F ss1 = gx*ifx;
-    F ss2 = (gx+1)*ifx;
-    F ss3 = (gx+2)*ifx;
-    F ss4 = (gx+3)*ifx;
-    F s5 = dy * ify;
-    sx.s0 = min((int)floor(ss1), src_cols-1);
-    sx.s1 = min((int)floor(ss2), src_cols-1);
-    sx.s2 = min((int)floor(ss3), src_cols-1);
-    sx.s3 = min((int)floor(ss4), src_cols-1);
-    sy = min((int)floor(s5), src_rows-1);
-
-    uchar4 val;
-    int4 pos = mad24((int4)sy, (int4)srcstep_in_pixel, sx+(int4)srcoffset_in_pixel);
-    val.s0 = src[pos.s0];
-    val.s1 = src[pos.s1];
-    val.s2 = src[pos.s2];
-    val.s3 = src[pos.s3];
-
-    //__global uchar4* d = (__global uchar4*)(dst + dstoffset_in_pixel + dy * dststep_in_pixel + gx);
-    //uchar4 dVal = *d;
-    pos = mad24(dy, dststep_in_pixel, gx+dstoffset_in_pixel);
-    pos.y++;
-    pos.z+=2;
-    pos.w+=3;
-
-        int con = (gx >= 0 && gx+3 < dst_cols && dy >= 0 && dy < dst_rows && (dstoffset_in_pixel&3)==0);
-    if(con)
-    {
-        *(__global uchar4*)(dst + pos.x)=val;
-    }
-    else
+    if (dx < dst_cols && dy < dst_rows)
     {
-        if(gx >= 0 && gx < dst_cols && dy >= 0 && dy < dst_rows)
-        {
-            dst[pos.x]=val.x;
-        }
-        if(gx+1 >= 0 && gx+1 < dst_cols && dy >= 0 && dy < dst_rows)
-        {
-            dst[pos.y]=val.y;
-        }
-        if(gx+2 >= 0 && gx+2 < dst_cols && dy >= 0 && dy < dst_rows)
-        {
-            dst[pos.z]=val.z;
-        }
-        if(gx+3 >= 0 && gx+3 < dst_cols && dy >= 0 && dy < dst_rows)
-        {
-            dst[pos.w]=val.w;
-        }
+        float s1 = dx * ifx, s2 = dy * ify;
+        int sx = min(convert_int_sat_rtn(s1), src_cols - 1);
+        int sy = min(convert_int_sat_rtn(s2), src_rows - 1);
+
+        int dst_index = mad24(dy, dst_step, dx + dst_offset);
+        int src_index = mad24(sy, src_step, sx + src_offset);
+
+        dst[dst_index] = src[src_index];
     }
 }
 
-__kernel void resizeNN_C4_D0(__global uchar4 * dst, __global uchar4 * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
-                     int src_cols, int src_rows, int dst_cols, int dst_rows, F ifx, F ify )
+#elif defined AREA
+
+#ifdef AREA_FAST
+
+__kernel void resizeAREA_FAST(__global T * dst, __global T * src,
+                         int dst_offset, int src_offset, int dst_step, int src_step,
+                         int src_cols, int src_rows, int dst_cols, int dst_rows, WT ifx, WT ify,
+                         __global const int * dmap_tab, __global const int * smap_tab)
 {
     int dx = get_global_id(0);
     int dy = get_global_id(1);
 
-    F s1 = dx*ifx;
-    F s2 = dy*ify;
-    int sx = fmin((float)floor(s1), (float)src_cols-1);
-    int sy = fmin((float)floor(s2), (float)src_rows-1);
-    int dpos = mad24(dy, dststep_in_pixel, dx + dstoffset_in_pixel);
-    int spos = mad24(sy, srcstep_in_pixel, sx + srcoffset_in_pixel);
+    if (dx < dst_cols && dy < dst_rows)
+    {
+        int dst_index = mad24(dy, dst_step, dst_offset + dx);
+
+        __global const int * xmap_tab = dmap_tab;
+        __global const int * ymap_tab = dmap_tab + dst_cols;
+        __global const int * sxmap_tab = smap_tab;
+        __global const int * symap_tab = smap_tab + XSCALE * dst_cols;
 
-    if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
-        dst[dpos] = src[spos];
+        int sx = xmap_tab[dx], sy = ymap_tab[dy];
+        WTV sum = (WTV)(0);
+
+        #pragma unroll
+        for (int y = 0; y < YSCALE; ++y)
+        {
+            int src_index = mad24(symap_tab[y + sy], src_step, src_offset);
+            #pragma unroll
+            for (int x = 0; x < XSCALE; ++x)
+                sum += convertToWTV(src[src_index + sxmap_tab[sx + x]]);
+        }
 
+        dst[dst_index] = convertToT(convertToWT2V(sum) * (WT2V)(SCALE));
+    }
 }
 
-__kernel void resizeNN_C1_D5(__global float * dst, __global float * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
-                     int src_cols, int src_rows, int dst_cols, int dst_rows, F ifx, F ify )
+#else
+
+__kernel void resizeAREA(__global T * dst, __global T * src,
+                         int dst_offset, int src_offset, int dst_step, int src_step,
+                         int src_cols, int src_rows, int dst_cols, int dst_rows, WT ifx, WT ify,
+                         __global const int * ofs_tab, __global const int * map_tab,
+                         __global const float * alpha_tab)
 {
     int dx = get_global_id(0);
     int dy = get_global_id(1);
 
-    F s1 = dx*ifx;
-    F s2 = dy*ify;
-    int sx = fmin((float)floor(s1), (float)src_cols-1);
-    int sy = fmin((float)floor(s2), (float)src_rows-1);
+    if (dx < dst_cols && dy < dst_rows)
+    {
+        int dst_index = mad24(dy, dst_step, dst_offset + dx);
 
-    int dpos = mad24(dy, dststep_in_pixel, dx + dstoffset_in_pixel);
-    int spos = mad24(sy, srcstep_in_pixel, sx + srcoffset_in_pixel);
-    if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
-        dst[dpos] = src[spos];
+        __global const int * xmap_tab = map_tab;
+        __global const int * ymap_tab = (__global const int *)(map_tab + (src_cols << 1));
+        __global const float * xalpha_tab = alpha_tab;
+        __global const float * yalpha_tab = (__global const float *)(alpha_tab + (src_cols << 1));
+        __global const int * xofs_tab = ofs_tab;
+        __global const int * yofs_tab = (__global const int *)(ofs_tab + dst_cols + 1);
 
-}
+        int xk0 = xofs_tab[dx], xk1 = xofs_tab[dx + 1];
+        int yk0 = yofs_tab[dy], yk1 = yofs_tab[dy + 1];
 
-__kernel void resizeNN_C4_D5(__global float4 * dst, __global float4 * src,
-                     int dstoffset_in_pixel, int srcoffset_in_pixel,int dststep_in_pixel, int srcstep_in_pixel,
-                     int src_cols, int src_rows, int dst_cols, int dst_rows, F ifx, F ify )
-{
-    int dx = get_global_id(0);
-    int dy = get_global_id(1);
-    F s1 = dx*ifx;
-    F s2 = dy*ify;
-    int s_col = floor(s1);
-    int s_row = floor(s2);
-    int sx = min(s_col, src_cols-1);
-    int sy = min(s_row, src_rows-1);
-    int dpos = mad24(dy, dststep_in_pixel, dx + dstoffset_in_pixel);
-    int spos = mad24(sy, srcstep_in_pixel, sx + srcoffset_in_pixel);
+        int sy0 = ymap_tab[yk0], sy1 = ymap_tab[yk1 - 1];
+        int sx0 = xmap_tab[xk0], sx1 = xmap_tab[xk1 - 1];
 
-    if(dx>=0 && dx<dst_cols && dy>=0 && dy<dst_rows)
-        dst[dpos] = src[spos];
+        WTV sum = (WTV)(0), buf;
+        int src_index = mad24(sy0, src_step, src_offset);
+
+        for (int sy = sy0, yk = yk0; sy <= sy1; ++sy, src_index += src_step, ++yk)
+        {
+            WTV beta = (WTV)(yalpha_tab[yk]);
+            buf = (WTV)(0);
+
+            for (int sx = sx0, xk = xk0; sx <= sx1; ++sx, ++xk)
+            {
+                WTV alpha = (WTV)(xalpha_tab[xk]);
+                buf += convertToWTV(src[src_index + sx]) * alpha;
+            }
+            sum += buf * beta;
+        }
 
+        dst[dst_index] = convertToT(sum);
+    }
 }
+
+#endif
+
+#endif

diff --git a/modules/ocl/src/opencl/imgproc_threshold.cl b/modules/ocl/src/opencl/imgproc_threshold.cl
index 400ac80..63e4102 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_threshold.cl
+++ b/modules/ocl/src/opencl/imgproc_threshold.cl
@@ -43,7 +43,7 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
 #ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
 #elif defined (cl_khr_fp64)
@@ -93,8 +93,8 @@ __kernel void threshold(__global const T * restrict src, int src_offset, int src
 #endif
         else
         {
-            T array[VECSIZE];
-            VSTOREN(vecValue, 0, array);
+            __attribute__(( aligned(sizeof(VT)) )) T array[VECSIZE];
+            *((VT*)array) = vecValue;
             #pragma unroll
             for (int i = 0; i < VECSIZE; ++i)
                 if (gx + i < max_index)

diff --git a/modules/ocl/src/opencl/imgproc_warpAffine.cl b/modules/ocl/src/opencl/imgproc_warpAffine.cl
index a5050bb..27f99e0 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_warpAffine.cl
+++ b/modules/ocl/src/opencl/imgproc_warpAffine.cl
@@ -47,11 +47,11 @@
 //warpAffine kernel
 //support data types: CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4, and three interpolation methods: NN, Linear, Cubic.
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 typedef double F;
 typedef double4 F4;

diff --git a/modules/ocl/src/opencl/imgproc_warpPerspective.cl b/modules/ocl/src/opencl/imgproc_warpPerspective.cl
index eee1c81..97f8664 100644 (file)
--- a/modules/ocl/src/opencl/imgproc_warpPerspective.cl
+++ b/modules/ocl/src/opencl/imgproc_warpPerspective.cl
@@ -47,11 +47,11 @@
 //wrapPerspective kernel
 //support data types: CV_8UC1, CV_8UC4, CV_32FC1, CV_32FC4, and three interpolation methods: NN, Linear, Cubic.
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 typedef double F;
 typedef double4 F4;

diff --git a/modules/ocl/src/opencl/kernel_stablesort_by_key.cl b/modules/ocl/src/opencl/kernel_stablesort_by_key.cl
index 2d38fbf..f8cc693 100644 (file)
--- a/modules/ocl/src/opencl/kernel_stablesort_by_key.cl
+++ b/modules/ocl/src/opencl/kernel_stablesort_by_key.cl
@@ -61,35 +61,6 @@
 #define my_comp(x,y) ((x) < (y))
 #endif
 
-///////////// parallel merge sort ///////////////
-// ported from https://github.com/HSA-Libraries/Bolt/blob/master/include/bolt/cl/stablesort_by_key_kernels.cl
-static uint lowerBoundLinear( global K_T* data, uint left, uint right, K_T searchVal)
-{
-    //  The values firstIndex and lastIndex get modified within the loop, narrowing down the potential sequence
-    uint firstIndex = left;
-    uint lastIndex = right;
-
-    //  This loops through [firstIndex, lastIndex)
-    //  Since firstIndex and lastIndex will be different for every thread depending on the nested branch,
-    //  this while loop will be divergent within a wavefront
-    while( firstIndex < lastIndex )
-    {
-        K_T dataVal = data[ firstIndex ];
-
-        //  This branch will create divergent wavefronts
-        if( my_comp( dataVal, searchVal ) )
-        {
-            firstIndex = firstIndex+1;
-        }
-        else
-        {
-            break;
-        }
-    }
-
-    return firstIndex;
-}
-
 //  This implements a binary search routine to look for an 'insertion point' in a sequence, denoted
 //  by a base pointer and left and right index for a particular candidate value.  The comparison operator is
 //  passed as a functor parameter my_comp

diff --git a/modules/ocl/src/opencl/kmeans_kernel.cl b/modules/ocl/src/opencl/kmeans_kernel.cl
index f62a08f..244d52c 100644 (file)
--- a/modules/ocl/src/opencl/kmeans_kernel.cl
+++ b/modules/ocl/src/opencl/kmeans_kernel.cl
@@ -44,81 +44,64 @@
 //
 //M*/
 
+static float distance_(__global const float * center, __global const float * src, int feature_length)
+{
+    float res = 0;
+    float4 v0, v1, v2;
+    int i = 0;
+
 #ifdef L1_DIST
-#  define DISTANCE(A, B) fabs((A) - (B))
-#elif defined L2SQR_DIST
-#  define DISTANCE(A, B) ((A) - (B)) * ((A) - (B))
-#else
-#  define DISTANCE(A, B) ((A) - (B)) * ((A) - (B))
+    float4 sum = (float4)(0.0f);
 #endif
 
-inline float dist(__global const float * center, __global const float * src, int feature_cols)
-{
-    float res = 0;
-    float4 tmp4;
-    int i;
-    for(i = 0; i < feature_cols / 4; i += 4, center += 4, src += 4)
+    for ( ; i <= feature_length - 4; i += 4)
     {
-        tmp4 = vload4(0, center) - vload4(0, src);
+        v0 = vload4(0, center + i);
+        v1 = vload4(0, src + i);
+        v2 = v1 - v0;
 #ifdef L1_DIST
-        tmp4 = fabs(tmp4);
+        v0 = fabs(v2);
+        sum += v0;
 #else
-        tmp4 *= tmp4;
+        res += dot(v2, v2);
 #endif
-        res += tmp4.x + tmp4.y + tmp4.z + tmp4.w;
     }
 
-    for(; i < feature_cols; ++i, ++center, ++src)
+#ifdef L1_DIST
+    res = sum.x + sum.y + sum.z + sum.w;
+#endif
+
+    for ( ; i < feature_length; ++i)
     {
-        res += DISTANCE(*src, *center);
+        float t0 = src[i];
+        float t1 = center[i];
+#ifdef L1_DIST
+        res += fabs(t0 - t1);
+#else
+        float t2 = t0 - t1;
+        res += t2 * t2;
+#endif
     }
+
     return res;
 }
 
-// to be distinguished with distanceToCenters in kmeans_kernel.cl
-__kernel void distanceToCenters(
-    __global const float *src,
-    __global const float *centers,
-#ifdef USE_INDEX
-    __global const int   *indices,
-#endif
-    __global int   *labels,
-    __global float *dists,
-    int feature_cols,
-    int src_step,
-    int centers_step,
-    int label_step,
-    int input_size,
-    int K,
-    int offset_src,
-    int offset_centers
-)
+__kernel void distanceToCenters(__global const float * src, __global const float * centers,
+                                __global float * dists, int feature_length,
+                                int src_step, int centers_step,
+                                int features_count, int centers_count,
+                                int src_offset, int centers_offset)
 {
     int gid = get_global_id(0);
-    float euDist, minval;
-    int minCentroid;
-    if(gid >= input_size)
-    {
-        return;
-    }
-    src += offset_src;
-    centers += offset_centers;
-#ifdef USE_INDEX
-    src += indices[gid] * src_step;
-#else
-    src += gid * src_step;
-#endif
-    minval = dist(centers, src, feature_cols);
-    minCentroid = 0;
-    for(int i = 1 ; i < K; i++)
+
+    if (gid < (features_count * centers_count))
     {
-        euDist = dist(centers + i * centers_step, src, feature_cols);
-        if(euDist < minval)
-        {
-            minval = euDist;
-            minCentroid = i;
-        }
+        int feature_index = gid / centers_count;
+        int center_index = gid % centers_count;
+
+        int center_idx = mad24(center_index, centers_step, centers_offset);
+        int src_idx = mad24(feature_index, src_step, src_offset);
+
+        dists[gid] = distance_(centers + center_idx, src + src_idx, feature_length);
     }
-    labels[gid * label_step] = minCentroid;
-    dists[gid] = minval;
 }

diff --git a/modules/ocl/src/opencl/knearest.cl b/modules/ocl/src/opencl/knearest.cl
index bc0ae89..85e2451 100644 (file)
--- a/modules/ocl/src/opencl/knearest.cl
+++ b/modules/ocl/src/opencl/knearest.cl
@@ -42,8 +42,13 @@
 // the use of this software, even if advised of the possibility of such damage.
 //
 //M*/
-#if defined (DOUBLE_SUPPORT)
+
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
+#endif
 #define TYPE double
 #else
 #define TYPE float

diff --git a/modules/ocl/src/opencl/match_template.cl b/modules/ocl/src/opencl/match_template.cl
index 8b63c3b..4d46d00 100644 (file)
--- a/modules/ocl/src/opencl/match_template.cl
+++ b/modules/ocl/src/opencl/match_template.cl
@@ -43,14 +43,12 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
-
 #define TYPE_IMAGE_SQSUM double
 #else
 #define TYPE_IMAGE_SQSUM float

diff --git a/modules/ocl/src/opencl/merge_mat.cl b/modules/ocl/src/opencl/merge_mat.cl
index 8b445c6..aea05ae 100644 (file)
--- a/modules/ocl/src/opencl/merge_mat.cl
+++ b/modules/ocl/src/opencl/merge_mat.cl
@@ -43,15 +43,19 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
-
+#endif
 
 ///////////////////////////////////////////////////////////////////////////////////////////////
 //////////////////////////////////optimized code using vector roi//////////////////////////
 ////////////vector fuction name format: merge_vector_C(channels number)D_(data type depth)//////
 ////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void merge_vector_C2_D0(__global uchar *mat_dst,  int dst_step,  int dst_offset,
                                  __global uchar *mat_src0, int src0_step, int src0_offset,
                                  __global uchar *mat_src1, int src1_step, int src1_offset,

diff --git a/modules/ocl/src/opencl/moments.cl b/modules/ocl/src/opencl/moments.cl
index 31c4c85..09c79c4 100644 (file)
--- a/modules/ocl/src/opencl/moments.cl
+++ b/modules/ocl/src/opencl/moments.cl
@@ -44,11 +44,11 @@
 //
 //M*/
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 typedef double T;
 #else

diff --git a/modules/ocl/src/opencl/operator_convertTo.cl b/modules/ocl/src/opencl/operator_convertTo.cl
index 85b562d..ca38bd5 100644 (file)
--- a/modules/ocl/src/opencl/operator_convertTo.cl
+++ b/modules/ocl/src/opencl/operator_convertTo.cl
@@ -35,8 +35,12 @@
 //
 
 #ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
 __kernel void convert_to(
         __global const srcT* restrict srcMat,

diff --git a/modules/ocl/src/opencl/operator_copyToM.cl b/modules/ocl/src/opencl/operator_copyToM.cl
index dcf5af9..69e1798 100644 (file)
--- a/modules/ocl/src/opencl/operator_copyToM.cl
+++ b/modules/ocl/src/opencl/operator_copyToM.cl
@@ -34,11 +34,11 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/operator_setTo.cl b/modules/ocl/src/opencl/operator_setTo.cl
index 8ac4803..20c5cf2 100644 (file)
--- a/modules/ocl/src/opencl/operator_setTo.cl
+++ b/modules/ocl/src/opencl/operator_setTo.cl
@@ -34,11 +34,11 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/operator_setToM.cl b/modules/ocl/src/opencl/operator_setToM.cl
index 8a489da..afaa2e6 100644 (file)
--- a/modules/ocl/src/opencl/operator_setToM.cl
+++ b/modules/ocl/src/opencl/operator_setToM.cl
@@ -34,11 +34,11 @@
 //
 //
 
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #endif
 

diff --git a/modules/ocl/src/opencl/pyrlk.cl b/modules/ocl/src/opencl/pyrlk.cl
index a7fc278..303d268 100644 (file)
--- a/modules/ocl/src/opencl/pyrlk.cl
+++ b/modules/ocl/src/opencl/pyrlk.cl
@@ -45,8 +45,6 @@
 //
 //M*/
 
-//#pragma OPENCL EXTENSION cl_amd_printf : enable
-
 #define        BUFFER  64
 #define        BUFFER2 BUFFER>>1
 #ifndef WAVE_SIZE

diff --git a/modules/ocl/src/opencl/split_mat.cl b/modules/ocl/src/opencl/split_mat.cl
index b9aa048..b52b3c2 100644 (file)
--- a/modules/ocl/src/opencl/split_mat.cl
+++ b/modules/ocl/src/opencl/split_mat.cl
@@ -38,9 +38,14 @@
 // the use of this software, even if advised of the possibility of such damage.
 //
 //M*/
-#if defined (DOUBLE_SUPPORT)
+
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
+#pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
+#endif
 
 #if DATA_DEPTH == 0
 #define BASE_TYPE uchar

diff --git a/modules/ocl/src/opencl/stereobm.cl b/modules/ocl/src/opencl/stereobm.cl
index 207bf00..0edccdb 100644 (file)
--- a/modules/ocl/src/opencl/stereobm.cl
+++ b/modules/ocl/src/opencl/stereobm.cl
@@ -260,7 +260,6 @@ static float CalcSums(__local float *cols, __local float *cols_cache, int winsz)
 {
     unsigned int cache = cols[0];
 
-#pragma unroll
     for(int i = 1; i <= winsz; i++)
         cache += cols[i];
 

diff --git a/modules/ocl/src/opencl/stereobp.cl b/modules/ocl/src/opencl/stereobp.cl
index ec02f82..4b5864f 100644 (file)
--- a/modules/ocl/src/opencl/stereobp.cl
+++ b/modules/ocl/src/opencl/stereobp.cl
@@ -45,13 +45,11 @@
 //M*/
 
 #if defined (DOUBLE_SUPPORT)
-
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
-
 #endif
 
 #ifdef T_FLOAT

diff --git a/modules/ocl/src/opencl/stereocsbp.cl b/modules/ocl/src/opencl/stereocsbp.cl
index 13a201c..72c1707 100644 (file)
--- a/modules/ocl/src/opencl/stereocsbp.cl
+++ b/modules/ocl/src/opencl/stereocsbp.cl
@@ -44,19 +44,10 @@
 //
 //M*/
 
-
-#ifndef FLT_MAX
-#define FLT_MAX  CL_FLT_MAX
-#endif
-
-#ifndef SHRT_MAX
-#define SHRT_MAX  CL_SHORT_MAX
-#endif
-
-
 ///////////////////////////////////////////////////////////////////////////////////////////////
 ////////////////////////////////////////get_first_k_initial_global//////////////////////////////
 //////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void get_first_k_initial_global_0(__global short *data_cost_selected_, __global short *selected_disp_pyr,
     __global short *ctemp, int h, int w, int nr_plane,
     int cmsg_step1, int cdisp_step1, int cndisp)
@@ -91,6 +82,7 @@ __kernel void get_first_k_initial_global_0(__global short *data_cost_selected_,
         }
     }
 }
+
 __kernel void get_first_k_initial_global_1(__global  float *data_cost_selected_, __global float *selected_disp_pyr,
     __global  float *ctemp, int h, int w, int nr_plane,
     int cmsg_step1, int cdisp_step1, int cndisp)
@@ -129,6 +121,7 @@ __kernel void get_first_k_initial_global_1(__global  float *data_cost_selected_,
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
 ///////////////////////////////////////////get_first_k_initial_local////////////////////////////////////
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void get_first_k_initial_local_0(__global  short *data_cost_selected_, __global short *selected_disp_pyr,
     __global  short *ctemp,int h, int w, int nr_plane,
     int cmsg_step1, int cdisp_step1, int cndisp)
@@ -248,6 +241,7 @@ __kernel void get_first_k_initial_local_1(__global float *data_cost_selected_, _
 ///////////////////////////////////////////////////////////////
 /////////////////////// init data cost ////////////////////////
 ///////////////////////////////////////////////////////////////
+
 inline float compute_3(__global uchar* left, __global uchar* right,
     float cdata_weight,  float cmax_data_term)
 {
@@ -257,6 +251,7 @@ inline float compute_3(__global uchar* left, __global uchar* right,
 
     return fmin(cdata_weight * (tr + tg + tb), cdata_weight * cmax_data_term);
 }
+
 inline float compute_1(__global uchar* left, __global uchar* right,
     float cdata_weight,  float cmax_data_term)
 {
@@ -316,6 +311,7 @@ __kernel void init_data_cost_0(__global short *ctemp, __global uchar *cleft, __g
         }
     }
 }
+
 __kernel void init_data_cost_1(__global float *ctemp, __global uchar *cleft, __global uchar *cright,
     int h, int w, int level, int channels,
     int cmsg_step1, float cdata_weight, float cmax_data_term, int cdisp_step1,
@@ -360,9 +356,11 @@ __kernel void init_data_cost_1(__global float *ctemp, __global uchar *cleft, __g
         }
     }
 }
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
 //////////////////////////////////init_data_cost_reduce//////////////////////////////////////////////////
 //////////////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void init_data_cost_reduce_0(__global short *ctemp, __global uchar *cleft, __global uchar *cright,
     __local float *smem, int level, int rows, int cols, int h, int winsz, int channels,
     int cndisp,int cimg_step, float cdata_weight, float cmax_data_term, int cth,
@@ -630,6 +628,7 @@ __kernel void init_data_cost_reduce_1(__global float *ctemp, __global uchar *cle
 ///////////////////////////////////////////////////////////////
 ////////////////////// compute data cost //////////////////////
 ///////////////////////////////////////////////////////////////
+
 __kernel void compute_data_cost_0(__global const short *selected_disp_pyr, __global short *data_cost_,
     __global uchar *cleft, __global uchar *cright,
     int h, int w, int level, int nr_plane, int channels,
@@ -680,6 +679,7 @@ __kernel void compute_data_cost_0(__global const short *selected_disp_pyr, __glo
         }
     }
 }
+
 __kernel void compute_data_cost_1(__global const float *selected_disp_pyr, __global float *data_cost_,
     __global uchar *cleft, __global uchar *cright,
     int h, int w, int level, int nr_plane, int channels,
@@ -729,9 +729,11 @@ __kernel void compute_data_cost_1(__global const float *selected_disp_pyr, __glo
         }
     }
 }
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
 ////////////////////////////////////////compute_data_cost_reduce//////////////////////////////////////////
 /////////////////////////////////////////////////////////////////////////////////////////////////////////
+
 __kernel void compute_data_cost_reduce_0(__global const short* selected_disp_pyr, __global short* data_cost_,
     __global uchar *cleft, __global uchar *cright,__local float *smem,
     int level, int rows, int cols, int h, int nr_plane,
@@ -1033,41 +1035,6 @@ static void get_first_k_element_increase_0(__global short* u_new, __global short
     }
 }
 
-static void get_first_k_element_increase_1(__global float *u_new, __global float *d_new, __global float *l_new,
-    __global float *r_new, __global const float *u_cur, __global const float *d_cur,
-    __global const float *l_cur, __global const float *r_cur,
-    __global float *data_cost_selected, __global float *disparity_selected_new,
-    __global float *data_cost_new, __global const float *data_cost_cur,
-    __global const float *disparity_selected_cur,
-    int nr_plane, int nr_plane2,
-    int cdisp_step1, int cdisp_step2)
-{
-    for(int i = 0; i < nr_plane; i++)
-    {
-        float minimum = FLT_MAX;
-        int id = 0;
-
-        for(int j = 0; j < nr_plane2; j++)
-        {
-            float cur = data_cost_new[j * cdisp_step1];
-            if(cur < minimum)
-            {
-                minimum = cur;
-                id = j;
-            }
-        }
-
-        data_cost_selected[i * cdisp_step1] = data_cost_cur[id * cdisp_step1];
-        disparity_selected_new[i * cdisp_step1] = disparity_selected_cur[id * cdisp_step2];
-
-        u_new[i * cdisp_step1] = u_cur[id * cdisp_step2];
-        d_new[i * cdisp_step1] = d_cur[id * cdisp_step2];
-        l_new[i * cdisp_step1] = l_cur[id * cdisp_step2];
-        r_new[i * cdisp_step1] = r_cur[id * cdisp_step2];
-        data_cost_new[id * cdisp_step1] = FLT_MAX;
-
-    }
-}
 __kernel void init_message_0(__global short *u_new_, __global short *d_new_, __global short *l_new_,
     __global short *r_new_, __global  short *u_cur_, __global const short *d_cur_,
     __global const short *l_cur_, __global const short *r_cur_, __global short *ctemp,
@@ -1118,6 +1085,7 @@ __kernel void init_message_0(__global short *u_new_, __global short *d_new_, __g
             cdisp_step1, cdisp_step2);
     }
 }
+
 __kernel void init_message_1(__global float *u_new_, __global float *d_new_, __global float *l_new_,
     __global float *r_new_, __global const float *u_cur_, __global const float *d_cur_,
     __global const float *l_cur_, __global const float *r_cur_, __global float *ctemp,

diff --git a/modules/ocl/src/opencl/svm.cl b/modules/ocl/src/opencl/svm.cl
index 36ae38e..32b8194 100644 (file)
--- a/modules/ocl/src/opencl/svm.cl
+++ b/modules/ocl/src/opencl/svm.cl
@@ -33,11 +33,12 @@
 // the use of this software, even if advised of the possibility of such damage.
 //
 //
-#if defined (DOUBLE_SUPPORT)
-#ifdef cl_khr_fp64
-#pragma OPENCL EXTENSION cl_khr_fp64:enable
-#elif defined (cl_amd_fp64)
+
+#ifdef DOUBLE_SUPPORT
+#ifdef cl_amd_fp64
 #pragma OPENCL EXTENSION cl_amd_fp64:enable
+#elif defined (cl_khr_fp64)
+#pragma OPENCL EXTENSION cl_khr_fp64:enable
 #endif
 #define TYPE double
 #else
@@ -53,7 +54,6 @@
 #else
 #define POW(X,Y) X
 #endif
-#define FLT_MAX   3.402823466e+38F
 #define MAX_VAL   (FLT_MAX*1e-3)
 
 __kernel void svm_linear(__global float* src, int src_step, __global float* src2, int src2_step, __global TYPE* dst, int dst_step, int src_rows, int src2_cols,
@@ -206,4 +206,4 @@ __kernel void svm_rbf(__global float* src, int src_step, __global float* src2, i
             dst[row * dst_step + col] = temp1;
         }
     }
-}
\ No newline at end of file
+}

diff --git a/modules/ocl/src/opencl/tvl1flow.cl b/modules/ocl/src/opencl/tvl1flow.cl
index 2787f00..6111a4a 100644 (file)
--- a/modules/ocl/src/opencl/tvl1flow.cl
+++ b/modules/ocl/src/opencl/tvl1flow.cl
@@ -44,7 +44,7 @@
 //M*/
 
 __kernel void centeredGradientKernel(__global const float* src, int src_col, int src_row, int src_step,
-__global float* dx, __global float* dy, int dx_step)
+                                     __global float* dx, __global float* dy, int dx_step)
 {
     int x = get_global_id(0);
     int y = get_global_id(1);
@@ -53,13 +53,6 @@ __global float* dx, __global float* dy, int dx_step)
     {
         int src_x1 = (x + 1) < (src_col -1)? (x + 1) : (src_col - 1);
         int src_x2 = (x - 1) > 0 ? (x -1) : 0;
-
-        //if(src[y * src_step + src_x1] == src[y * src_step+ src_x2])
-        //{
-        //    printf("y = %d\n", y);
-        //    printf("src_x1 = %d\n", src_x1);
-        //    printf("src_x2 = %d\n", src_x2);
-        //}
         dx[y * dx_step+ x] = 0.5f * (src[y * src_step + src_x1] - src[y * src_step+ src_x2]);
 
         int src_y1 = (y+1) < (src_row - 1) ? (y + 1) : (src_row - 1);
@@ -97,24 +90,24 @@ __kernel void warpBackwardKernel(__global const float* I0, int I0_step, int I0_c
     int u2_offset_x,
     int u2_offset_y)
 {
-    const int x = get_global_id(0);
-    const int y = get_global_id(1);
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
     if(x < I0_col&&y < I0_row)
     {
-        //const float u1Val = u1(y, x);
-        const float u1Val = u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
-        //const float u2Val = u2(y, x);
-        const float u2Val = u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
+        //float u1Val = u1(y, x);
+        float u1Val = u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
+        //float u2Val = u2(y, x);
+        float u2Val = u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
 
-        const float wx = x + u1Val;
-        const float wy = y + u2Val;
+        float wx = x + u1Val;
+        float wy = y + u2Val;
 
-        const int xmin = ceil(wx - 2.0f);
-        const int xmax = floor(wx + 2.0f);
+        int xmin = ceil(wx - 2.0f);
+        int xmax = floor(wx + 2.0f);
 
-        const int ymin = ceil(wy - 2.0f);
-        const int ymax = floor(wy + 2.0f);
+        int ymin = ceil(wy - 2.0f);
+        int ymax = floor(wy + 2.0f);
 
         float sum  = 0.0f;
         float sumx = 0.0f;
@@ -126,7 +119,7 @@ __kernel void warpBackwardKernel(__global const float* I0, int I0_step, int I0_c
         {
             for (int cx = xmin; cx <= xmax; ++cx)
             {
-                const float w = bicubicCoeff(wx - cx) * bicubicCoeff(wy - cy);
+                float w = bicubicCoeff(wx - cx) * bicubicCoeff(wy - cy);
 
                 //sum  += w * tex2D(tex_I1 , cx, cy);
                 int2 cood = (int2)(cx, cy);
@@ -140,30 +133,30 @@ __kernel void warpBackwardKernel(__global const float* I0, int I0_step, int I0_c
             }
         }
 
-        const float coeff = 1.0f / wsum;
+        float coeff = 1.0f / wsum;
 
-        const float I1wVal  = sum  * coeff;
-        const float I1wxVal = sumx * coeff;
-        const float I1wyVal = sumy * coeff;
+        float I1wVal  = sum  * coeff;
+        float I1wxVal = sumx * coeff;
+        float I1wyVal = sumy * coeff;
 
         I1w[y * I1w_step + x]  = I1wVal;
         I1wx[y * I1w_step + x] = I1wxVal;
         I1wy[y * I1w_step + x] = I1wyVal;
 
-        const float Ix2 = I1wxVal * I1wxVal;
-        const float Iy2 = I1wyVal * I1wyVal;
+        float Ix2 = I1wxVal * I1wxVal;
+        float Iy2 = I1wyVal * I1wyVal;
 
         // store the |Grad(I1)|^2
         grad[y * I1w_step + x] = Ix2 + Iy2;
 
         // compute the constant part of the rho function
-        const float I0Val = I0[y * I0_step + x];
+        float I0Val = I0[y * I0_step + x];
         rho[y * I1w_step + x] = I1wVal - I1wxVal * u1Val - I1wyVal * u2Val - I0Val;
     }
 
 }
 
-static float readImage(__global const float *image,  const int x,  const int y,  const int rows,  const int cols, const int elemCntPerRow)
+static float readImage(__global float *image,  int x,  int y,  int rows,  int cols, int elemCntPerRow)
 {
     int i0 = clamp(x, 0, cols - 1);
     int j0 = clamp(y, 0, rows - 1);
@@ -185,24 +178,24 @@ __kernel void warpBackwardKernelNoImage2d(__global const float* I0, int I0_step,
     int I1_step,
     int I1x_step)
 {
-    const int x = get_global_id(0);
-    const int y = get_global_id(1);
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
     if(x < I0_col&&y < I0_row)
     {
-        //const float u1Val = u1(y, x);
-        const float u1Val = u1[y * u1_step + x];
-        //const float u2Val = u2(y, x);
-        const float u2Val = u2[y * u2_step + x];
+        //float u1Val = u1(y, x);
+        float u1Val = u1[y * u1_step + x];
+        //float u2Val = u2(y, x);
+        float u2Val = u2[y * u2_step + x];
 
-        const float wx = x + u1Val;
-        const float wy = y + u2Val;
+        float wx = x + u1Val;
+        float wy = y + u2Val;
 
-        const int xmin = ceil(wx - 2.0f);
-        const int xmax = floor(wx + 2.0f);
+        int xmin = ceil(wx - 2.0f);
+        int xmax = floor(wx + 2.0f);
 
-        const int ymin = ceil(wy - 2.0f);
-        const int ymax = floor(wy + 2.0f);
+        int ymin = ceil(wy - 2.0f);
+        int ymax = floor(wy + 2.0f);
 
         float sum  = 0.0f;
         float sumx = 0.0f;
@@ -213,7 +206,7 @@ __kernel void warpBackwardKernelNoImage2d(__global const float* I0, int I0_step,
         {
             for (int cx = xmin; cx <= xmax; ++cx)
             {
-                const float w = bicubicCoeff(wx - cx) * bicubicCoeff(wy - cy);
+                float w = bicubicCoeff(wx - cx) * bicubicCoeff(wy - cy);
 
                 int2 cood = (int2)(cx, cy);
                 sum += w * readImage(tex_I1, cood.x, cood.y, I0_col, I0_row, I1_step);
@@ -223,24 +216,24 @@ __kernel void warpBackwardKernelNoImage2d(__global const float* I0, int I0_step,
             }
         }
 
-        const float coeff = 1.0f / wsum;
+        float coeff = 1.0f / wsum;
 
-        const float I1wVal  = sum  * coeff;
-        const float I1wxVal = sumx * coeff;
-        const float I1wyVal = sumy * coeff;
+        float I1wVal  = sum  * coeff;
+        float I1wxVal = sumx * coeff;
+        float I1wyVal = sumy * coeff;
 
         I1w[y * I1w_step + x]  = I1wVal;
         I1wx[y * I1w_step + x] = I1wxVal;
         I1wy[y * I1w_step + x] = I1wyVal;
 
-        const float Ix2 = I1wxVal * I1wxVal;
-        const float Iy2 = I1wyVal * I1wyVal;
+        float Ix2 = I1wxVal * I1wxVal;
+        float Iy2 = I1wyVal * I1wyVal;
 
         // store the |Grad(I1)|^2
         grad[y * I1w_step + x] = Ix2 + Iy2;
 
         // compute the constant part of the rho function
-        const float I0Val = I0[y * I0_step + x];
+        float I0Val = I0[y * I0_step + x];
         rho[y * I1w_step + x] = I1wVal - I1wxVal * u1Val - I1wyVal * u2Val - I0Val;
     }
 
@@ -253,38 +246,35 @@ __kernel void estimateDualVariablesKernel(__global const float* u1, int u1_col,
     __global float* p12,
     __global float* p21,
     __global float* p22,
-    const float taut,
+    float taut,
     int u2_step,
     int u1_offset_x,
     int u1_offset_y,
     int u2_offset_x,
     int u2_offset_y)
 {
-
-    //const int x = blockIdx.x * blockDim.x + threadIdx.x;
-    //const int y = blockIdx.y * blockDim.y + threadIdx.y;
-    const int x = get_global_id(0);
-    const int y = get_global_id(1);
+    int x = get_global_id(0);
+    int y = get_global_id(1);
 
     if(x < u1_col && y < u1_row)
     {
         int src_x1 = (x + 1) < (u1_col - 1) ? (x + 1) : (u1_col - 1);
-        const float u1x = u1[(y + u1_offset_y) * u1_step + src_x1 + u1_offset_x] - u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
+        float u1x = u1[(y + u1_offset_y) * u1_step + src_x1 + u1_offset_x] - u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
 
         int src_y1 = (y + 1) < (u1_row - 1) ? (y + 1) : (u1_row - 1);
-        const float u1y = u1[(src_y1 + u1_offset_y) * u1_step + x + u1_offset_x] - u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
+        float u1y = u1[(src_y1 + u1_offset_y) * u1_step + x + u1_offset_x] - u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
 
         int src_x2 = (x + 1) < (u1_col - 1) ? (x + 1) : (u1_col - 1);
-        const float u2x = u2[(y + u2_offset_y) * u2_step + src_x2 + u2_offset_x] - u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
+        float u2x = u2[(y + u2_offset_y) * u2_step + src_x2 + u2_offset_x] - u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
 
         int src_y2 = (y + 1) <  (u1_row - 1) ? (y + 1) : (u1_row - 1);
-        const float u2y = u2[(src_y2 + u2_offset_y) * u2_step + x + u2_offset_x] - u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
+        float u2y = u2[(src_y2 + u2_offset_y) * u2_step + x + u2_offset_x] - u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
 
-        const float g1 = hypot(u1x, u1y);
-        const float g2 = hypot(u2x, u2y);
+        float g1 = hypot(u1x, u1y);
+        float g2 = hypot(u2x, u2y);
 
-        const float ng1 = 1.0f + taut * g1;
-        const float ng2 = 1.0f + taut * g2;
+        float ng1 = 1.0f + taut * g1;
+        float ng2 = 1.0f + taut * g2;
 
         p11[y * p11_step + x] = (p11[y * p11_step + x] + taut * u1x) / ng1;
         p12[y * p11_step + x] = (p12[y * p11_step + x] + taut * u1y) / ng1;
@@ -299,8 +289,8 @@ static float divergence(__global const float* v1, __global const float* v2, int
 
     if (x > 0 && y > 0)
     {
-        const float v1x = v1[y * v1_step + x] - v1[y * v1_step + x - 1];
-        const float v2y = v2[y * v2_step + x] - v2[(y - 1) * v2_step + x];
+        float v1x = v1[y * v1_step + x] - v1[y * v1_step + x - 1];
+        float v2y = v2[y * v2_step + x] - v2[(y - 1) * v2_step + x];
         return v1x + v2y;
     }
     else
@@ -328,30 +318,25 @@ __kernel void estimateUKernel(__global const float* I1wx, int I1wx_col, int I1wx
     __global const float* p22, /*int p22_step,*/
     __global float* u1, int u1_step,
     __global float* u2,
-    __global float* error, const float l_t, const float theta, int u2_step,
+    __global float* error, float l_t, float theta, int u2_step,
     int u1_offset_x,
     int u1_offset_y,
     int u2_offset_x,
     int u2_offset_y,
     char calc_error)
 {
-
-    //const int x = blockIdx.x * blockDim.x + threadIdx.x;
-    //const int y = blockIdx.y * blockDim.y + threadIdx.y;
-
     int x = get_global_id(0);
     int y = get_global_id(1);
 
-
     if(x < I1wx_col && y < I1wx_row)
     {
-        const float I1wxVal = I1wx[y * I1wx_step + x];
-        const float I1wyVal = I1wy[y * I1wx_step + x];
-        const float gradVal = grad[y * I1wx_step + x];
-        const float u1OldVal = u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
-        const float u2OldVal = u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
+        float I1wxVal = I1wx[y * I1wx_step + x];
+        float I1wyVal = I1wy[y * I1wx_step + x];
+        float gradVal = grad[y * I1wx_step + x];
+        float u1OldVal = u1[(y + u1_offset_y) * u1_step + x + u1_offset_x];
+        float u2OldVal = u2[(y + u2_offset_y) * u2_step + x + u2_offset_x];
 
-        const float rho = rho_c[y * I1wx_step + x] + (I1wxVal * u1OldVal + I1wyVal * u2OldVal);
+        float rho = rho_c[y * I1wx_step + x] + (I1wxVal * u1OldVal + I1wyVal * u2OldVal);
 
         // estimate the values of the variable (v1, v2) (thresholding operator TH)
 
@@ -370,31 +355,31 @@ __kernel void estimateUKernel(__global const float* I1wx, int I1wx_col, int I1wx
         }
         else if (gradVal > 1.192092896e-07f)
         {
-            const float fi = -rho / gradVal;
+            float fi = -rho / gradVal;
             d1 = fi * I1wxVal;
             d2 = fi * I1wyVal;
         }
 
-        const float v1 = u1OldVal + d1;
-        const float v2 = u2OldVal + d2;
+        float v1 = u1OldVal + d1;
+        float v2 = u2OldVal + d2;
 
         // compute the divergence of the dual variable (p1, p2)
 
-        const float div_p1 = divergence(p11, p12, y, x, I1wx_step, I1wx_step);
-        const float div_p2 = divergence(p21, p22, y, x, I1wx_step, I1wx_step);
+        float div_p1 = divergence(p11, p12, y, x, I1wx_step, I1wx_step);
+        float div_p2 = divergence(p21, p22, y, x, I1wx_step, I1wx_step);
 
         // estimate the values of the optical flow (u1, u2)
 
-        const float u1NewVal = v1 + theta * div_p1;
-        const float u2NewVal = v2 + theta * div_p2;
+        float u1NewVal = v1 + theta * div_p1;
+        float u2NewVal = v2 + theta * div_p2;
 
         u1[(y + u1_offset_y) * u1_step + x + u1_offset_x] = u1NewVal;
         u2[(y + u2_offset_y) * u2_step + x + u2_offset_x] = u2NewVal;
 
         if(calc_error)
         {
-            const float n1 = (u1OldVal - u1NewVal) * (u1OldVal - u1NewVal);
-            const float n2 = (u2OldVal - u2NewVal) * (u2OldVal - u2NewVal);
+            float n1 = (u1OldVal - u1NewVal) * (u1OldVal - u1NewVal);
+            float n2 = (u2OldVal - u2NewVal) * (u2OldVal - u2NewVal);
             error[y * I1wx_step + x] = n1 + n2;
         }
     }

diff --git a/modules/ocl/src/safe_call.hpp b/modules/ocl/src/safe_call.hpp
index f772e1b..679c89d 100644 (file)
--- a/modules/ocl/src/safe_call.hpp
+++ b/modules/ocl/src/safe_call.hpp
@@ -48,13 +48,8 @@
 
 #include "opencv2/ocl/cl_runtime/cl_runtime.hpp"
 
-#if defined(__GNUC__)
-#define openCLSafeCall(expr)  ___openCLSafeCall(expr, __FILE__, __LINE__, __func__)
-#define openCLVerifyCall(res) ___openCLSafeCall(res, __FILE__, __LINE__, __func__)
-#else /* defined(__OPENCLCC__) || defined(__MSVC__) */
-#define openCLSafeCall(expr)  ___openCLSafeCall(expr, __FILE__, __LINE__)
-#define openCLVerifyCall(res) ___openCLSafeCall(res, __FILE__, __LINE__)
-#endif
+#define openCLSafeCall(expr)  ___openCLSafeCall(expr, __FILE__, __LINE__, CV_Func)
+#define openCLVerifyCall(res) ___openCLSafeCall(res, __FILE__, __LINE__, CV_Func)
 
 
 namespace cv

diff --git a/modules/ocl/src/split_merge.cpp b/modules/ocl/src/split_merge.cpp
index 60a27a5..400a055 100644 (file)
--- a/modules/ocl/src/split_merge.cpp
+++ b/modules/ocl/src/split_merge.cpp
@@ -250,7 +250,8 @@ namespace cv
                         && devInfo.deviceType == CVCL_DEVICE_TYPE_CPU
                         && devInfo.platform->platformVendor.find("Intel") != std::string::npos
                         && (devInfo.deviceVersion.find("Build 56860") != std::string::npos
-                            || devInfo.deviceVersion.find("Build 76921") != std::string::npos))
+                            || devInfo.deviceVersion.find("Build 76921") != std::string::npos
+                            || devInfo.deviceVersion.find("Build 78712") != std::string::npos))
                     build_options += " -D BYPASS_VSTORE=true";
 
                 size_t globalThreads[3] = { divUp(src.cols, VEC_SIZE), src.rows, 1 };

diff --git a/modules/ocl/test/main.cpp b/modules/ocl/test/main.cpp
index d76fa84..992aeb8 100644 (file)
--- a/modules/ocl/test/main.cpp
+++ b/modules/ocl/test/main.cpp
@@ -42,15 +42,17 @@
 
 #include "test_precomp.hpp"
 
-#define DUMP_INFO_STDOUT(propertyDisplayName, propertyValue) \
+#define DUMP_PROPERTY_XML(propertyName, propertyValue) \
     do { \
-        std::cout << (propertyDisplayName) << ": " << (propertyValue) << std::endl; \
+        std::stringstream ssName, ssValue;\
+        ssName << propertyName;\
+        ssValue << propertyValue; \
+        ::testing::Test::RecordProperty(ssName.str(), ssValue.str()); \
     } while (false)
 
-#define DUMP_INFO_XML(propertyXMLName, propertyValue) \
+#define DUMP_MESSAGE_STDOUT(msg) \
     do { \
-        std::stringstream ss; ss << propertyValue; \
-        ::testing::Test::RecordProperty((propertyXMLName), ss.str()); \
+        std::cout << msg << std::endl; \
     } while (false)
 
 #include "opencv2/ocl/private/opencl_dumpinfo.hpp"

diff --git a/modules/ocl/test/test_color.cpp b/modules/ocl/test/test_color.cpp
index 4b6a6e4..7c4fe39 100644 (file)
--- a/modules/ocl/test/test_color.cpp
+++ b/modules/ocl/test/test_color.cpp
@@ -46,9 +46,8 @@
 #include "test_precomp.hpp"
 #ifdef HAVE_OPENCL
 
-namespace
-{
 using namespace testing;
+using namespace cv;
 
 ///////////////////////////////////////////////////////////////////////////////////////////////////////
 // cvtColor
@@ -59,20 +58,20 @@ PARAM_TEST_CASE(CvtColor, MatDepth, bool)
     bool use_roi;
 
     // src mat
-    cv::Mat src1;
-    cv::Mat dst1;
+    Mat src;
+    Mat dst;
 
     // src mat with roi
-    cv::Mat src1_roi;
-    cv::Mat dst1_roi;
+    Mat src_roi;
+    Mat dst_roi;
 
     // ocl dst mat for testing
-    cv::ocl::oclMat gsrc1_whole;
-    cv::ocl::oclMat gdst1_whole;
+    ocl::oclMat gsrc_whole;
+    ocl::oclMat gdst_whole;
 
     // ocl mat with roi
-    cv::ocl::oclMat gsrc1_roi;
-    cv::ocl::oclMat gdst1_roi;
+    ocl::oclMat gsrc_roi;
+    ocl::oclMat gdst_roi;
 
     virtual void SetUp()
     {
@@ -87,99 +86,188 @@ PARAM_TEST_CASE(CvtColor, MatDepth, bool)
 
         Size roiSize = randomSize(1, MAX_VALUE);
         Border srcBorder = randomBorder(0, use_roi ? MAX_VALUE : 0);
-        randomSubMat(src1, src1_roi, roiSize, srcBorder, srcType, 2, 100);
+        randomSubMat(src, src_roi, roiSize, srcBorder, srcType, 2, 100);
 
-        Border dst1Border = randomBorder(0, use_roi ? MAX_VALUE : 0);
-        randomSubMat(dst1, dst1_roi, roiSize, dst1Border, dstType, 5, 16);
+        Border dstBorder = randomBorder(0, use_roi ? MAX_VALUE : 0);
+        randomSubMat(dst, dst_roi, roiSize, dstBorder, dstType, 5, 16);
 
-        generateOclMat(gsrc1_whole, gsrc1_roi, src1, roiSize, srcBorder);
-        generateOclMat(gdst1_whole, gdst1_roi, dst1, roiSize, dst1Border);
+        generateOclMat(gsrc_whole, gsrc_roi, src, roiSize, srcBorder);
+        generateOclMat(gdst_whole, gdst_roi, dst, roiSize, dstBorder);
     }
 
-    void Near(double threshold = 1e-3)
+    void Near(double threshold)
     {
-        EXPECT_MAT_NEAR(dst1, gdst1_whole, threshold);
-        EXPECT_MAT_NEAR(dst1_roi, gdst1_roi, threshold);
+        Mat whole, roi;
+        gdst_whole.download(whole);
+        gdst_roi.download(roi);
+
+        EXPECT_MAT_NEAR(dst, whole, threshold);
+        EXPECT_MAT_NEAR(dst_roi, roi, threshold);
     }
 
-    void doTest(int channelsIn, int channelsOut, int code)
+    void doTest(int channelsIn, int channelsOut, int code, double threshold = 1e-3)
     {
         for (int j = 0; j < LOOP_TIMES; j++)
         {
             random_roi(channelsIn, channelsOut);
 
-            cv::cvtColor(src1_roi, dst1_roi, code);
-            cv::ocl::cvtColor(gsrc1_roi, gdst1_roi, code);
+            cvtColor(src_roi, dst_roi, code, channelsOut);
+            ocl::cvtColor(gsrc_roi, gdst_roi, code, channelsOut);
 
-            Near();
+            Near(threshold);
         }
     }
 };
 
-#define CVTCODE(name) cv::COLOR_ ## name
+#define CVTCODE(name) COLOR_ ## name
+
+// RGB[A] <-> BGR[A]
+
+OCL_TEST_P(CvtColor, BGR2BGRA) { doTest(3, 4, CVTCODE(BGR2BGRA)); }
+OCL_TEST_P(CvtColor, RGB2RGBA) { doTest(3, 4, CVTCODE(RGB2RGBA)); }
+OCL_TEST_P(CvtColor, BGRA2BGR) { doTest(4, 3, CVTCODE(BGRA2BGR)); }
+OCL_TEST_P(CvtColor, RGBA2RGB) { doTest(4, 3, CVTCODE(RGBA2RGB)); }
+OCL_TEST_P(CvtColor, BGR2RGBA) { doTest(3, 4, CVTCODE(BGR2RGBA)); }
+OCL_TEST_P(CvtColor, RGB2BGRA) { doTest(3, 4, CVTCODE(RGB2BGRA)); }
+OCL_TEST_P(CvtColor, RGBA2BGR) { doTest(4, 3, CVTCODE(RGBA2BGR)); }
+OCL_TEST_P(CvtColor, BGRA2RGB) { doTest(4, 3, CVTCODE(BGRA2RGB)); }
+OCL_TEST_P(CvtColor, BGR2RGB) { doTest(3, 3, CVTCODE(BGR2RGB)); }
+OCL_TEST_P(CvtColor, RGB2BGR) { doTest(3, 3, CVTCODE(RGB2BGR)); }
+OCL_TEST_P(CvtColor, BGRA2RGBA) { doTest(4, 4, CVTCODE(BGRA2RGBA)); }
+OCL_TEST_P(CvtColor, RGBA2BGRA) { doTest(4, 4, CVTCODE(RGBA2BGRA)); }
+
+// RGB <-> Gray
+
+OCL_TEST_P(CvtColor, RGB2GRAY) { doTest(3, 1, CVTCODE(RGB2GRAY)); }
+OCL_TEST_P(CvtColor, GRAY2RGB) { doTest(1, 3, CVTCODE(GRAY2RGB)); }
+OCL_TEST_P(CvtColor, BGR2GRAY) { doTest(3, 1, CVTCODE(BGR2GRAY)); }
+OCL_TEST_P(CvtColor, GRAY2BGR) { doTest(1, 3, CVTCODE(GRAY2BGR)); }
+OCL_TEST_P(CvtColor, RGBA2GRAY) { doTest(4, 1, CVTCODE(RGBA2GRAY)); }
+OCL_TEST_P(CvtColor, GRAY2RGBA) { doTest(1, 4, CVTCODE(GRAY2RGBA)); }
+OCL_TEST_P(CvtColor, BGRA2GRAY) { doTest(4, 1, CVTCODE(BGRA2GRAY)); }
+OCL_TEST_P(CvtColor, GRAY2BGRA) { doTest(1, 4, CVTCODE(GRAY2BGRA)); }
+
+// RGB <-> YUV
+
+OCL_TEST_P(CvtColor, RGB2YUV) { doTest(3, 3, CVTCODE(RGB2YUV)); }
+OCL_TEST_P(CvtColor, BGR2YUV) { doTest(3, 3, CVTCODE(BGR2YUV)); }
+OCL_TEST_P(CvtColor, RGBA2YUV) { doTest(4, 3, CVTCODE(RGB2YUV)); }
+OCL_TEST_P(CvtColor, BGRA2YUV) { doTest(4, 3, CVTCODE(BGR2YUV)); }
+OCL_TEST_P(CvtColor, YUV2RGB) { doTest(3, 3, CVTCODE(YUV2RGB)); }
+OCL_TEST_P(CvtColor, YUV2BGR) { doTest(3, 3, CVTCODE(YUV2BGR)); }
+OCL_TEST_P(CvtColor, YUV2RGBA) { doTest(3, 4, CVTCODE(YUV2RGB)); }
+OCL_TEST_P(CvtColor, YUV2BGRA) { doTest(3, 4, CVTCODE(YUV2BGR)); }
+
+// RGB <-> YCrCb
+
+OCL_TEST_P(CvtColor, RGB2YCrCb) { doTest(3, 3, CVTCODE(RGB2YCrCb)); }
+OCL_TEST_P(CvtColor, BGR2YCrCb) { doTest(3, 3, CVTCODE(BGR2YCrCb)); }
+OCL_TEST_P(CvtColor, RGBA2YCrCb) { doTest(4, 3, CVTCODE(RGB2YCrCb)); }
+OCL_TEST_P(CvtColor, BGRA2YCrCb) { doTest(4, 3, CVTCODE(BGR2YCrCb)); }
+OCL_TEST_P(CvtColor, YCrCb2RGB) { doTest(3, 3, CVTCODE(YCrCb2RGB)); }
+OCL_TEST_P(CvtColor, YCrCb2BGR) { doTest(3, 3, CVTCODE(YCrCb2BGR)); }
+OCL_TEST_P(CvtColor, YCrCb2RGBA) { doTest(3, 4, CVTCODE(YCrCb2RGB)); }
+OCL_TEST_P(CvtColor, YCrCb2BGRA) { doTest(3, 4, CVTCODE(YCrCb2BGR)); }
+
+// RGB <-> XYZ
+
+OCL_TEST_P(CvtColor, RGB2XYZ) { doTest(3, 3, CVTCODE(RGB2XYZ)); }
+OCL_TEST_P(CvtColor, BGR2XYZ) { doTest(3, 3, CVTCODE(BGR2XYZ)); }
+OCL_TEST_P(CvtColor, RGBA2XYZ) { doTest(4, 3, CVTCODE(RGB2XYZ)); }
+OCL_TEST_P(CvtColor, BGRA2XYZ) { doTest(4, 3, CVTCODE(BGR2XYZ)); }
+
+OCL_TEST_P(CvtColor, XYZ2RGB) { doTest(3, 3, CVTCODE(XYZ2RGB)); }
+OCL_TEST_P(CvtColor, XYZ2BGR) { doTest(3, 3, CVTCODE(XYZ2BGR)); }
+OCL_TEST_P(CvtColor, XYZ2RGBA) { doTest(3, 4, CVTCODE(XYZ2RGB)); }
+OCL_TEST_P(CvtColor, XYZ2BGRA) { doTest(3, 4, CVTCODE(XYZ2BGR)); }
+
+// RGB <-> HSV
+
+typedef CvtColor CvtColor8u32f;
+
+OCL_TEST_P(CvtColor8u32f, RGB2HSV) { doTest(3, 3, CVTCODE(RGB2HSV)); }
+OCL_TEST_P(CvtColor8u32f, BGR2HSV) { doTest(3, 3, CVTCODE(BGR2HSV)); }
+OCL_TEST_P(CvtColor8u32f, RGBA2HSV) { doTest(4, 3, CVTCODE(RGB2HSV)); }
+OCL_TEST_P(CvtColor8u32f, BGRA2HSV) { doTest(4, 3, CVTCODE(BGR2HSV)); }
+
+OCL_TEST_P(CvtColor8u32f, RGB2HSV_FULL) { doTest(3, 3, CVTCODE(RGB2HSV_FULL)); }
+OCL_TEST_P(CvtColor8u32f, BGR2HSV_FULL) { doTest(3, 3, CVTCODE(BGR2HSV_FULL)); }
+OCL_TEST_P(CvtColor8u32f, RGBA2HSV_FULL) { doTest(4, 3, CVTCODE(RGB2HSV_FULL)); }
+OCL_TEST_P(CvtColor8u32f, BGRA2HSV_FULL) { doTest(4, 3, CVTCODE(BGR2HSV_FULL)); }
+
+OCL_TEST_P(CvtColor8u32f, HSV2RGB) { doTest(3, 3, CVTCODE(HSV2RGB), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2BGR) { doTest(3, 3, CVTCODE(HSV2BGR), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2RGBA) { doTest(3, 4, CVTCODE(HSV2RGB), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2BGRA) { doTest(3, 4, CVTCODE(HSV2BGR), depth == CV_8U ? 1 : 4e-1); }
+
+OCL_TEST_P(CvtColor8u32f, HSV2RGB_FULL) { doTest(3, 3, CVTCODE(HSV2RGB_FULL), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2BGR_FULL) { doTest(3, 3, CVTCODE(HSV2BGR_FULL), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2RGBA_FULL) { doTest(3, 4, CVTCODE(HSV2BGR_FULL), depth == CV_8U ? 1 : 4e-1); }
+OCL_TEST_P(CvtColor8u32f, HSV2BGRA_FULL) { doTest(3, 4, CVTCODE(HSV2BGR_FULL), depth == CV_8U ? 1 : 4e-1); }
+
+// RGB <-> HLS
+
+OCL_TEST_P(CvtColor8u32f, RGB2HLS) { doTest(3, 3, CVTCODE(RGB2HLS), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, BGR2HLS) { doTest(3, 3, CVTCODE(BGR2HLS), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, RGBA2HLS) { doTest(4, 3, CVTCODE(RGB2HLS), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, BGRA2HLS) { doTest(4, 3, CVTCODE(BGR2HLS), depth == CV_8U ? 1 : 1e-3); }
+
+OCL_TEST_P(CvtColor8u32f, RGB2HLS_FULL) { doTest(3, 3, CVTCODE(RGB2HLS_FULL), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, BGR2HLS_FULL) { doTest(3, 3, CVTCODE(BGR2HLS_FULL), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, RGBA2HLS_FULL) { doTest(4, 3, CVTCODE(RGB2HLS_FULL), depth == CV_8U ? 1 : 1e-3); }
+OCL_TEST_P(CvtColor8u32f, BGRA2HLS_FULL) { doTest(4, 3, CVTCODE(BGR2HLS_FULL), depth == CV_8U ? 1 : 1e-3); }
+
+OCL_TEST_P(CvtColor8u32f, HLS2RGB) { doTest(3, 3, CVTCODE(HLS2RGB), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2BGR) { doTest(3, 3, CVTCODE(HLS2BGR), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2RGBA) { doTest(3, 4, CVTCODE(HLS2RGB), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2BGRA) { doTest(3, 4, CVTCODE(HLS2BGR), 1); }
+
+OCL_TEST_P(CvtColor8u32f, HLS2RGB_FULL) { doTest(3, 3, CVTCODE(HLS2RGB_FULL), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2BGR_FULL) { doTest(3, 3, CVTCODE(HLS2BGR_FULL), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2RGBA_FULL) { doTest(3, 4, CVTCODE(HLS2RGB_FULL), 1); }
+OCL_TEST_P(CvtColor8u32f, HLS2BGRA_FULL) { doTest(3, 4, CVTCODE(HLS2BGR_FULL), 1); }
+
+// RGB5x5 <-> RGB
+
+typedef CvtColor CvtColor8u;
+
+OCL_TEST_P(CvtColor8u, BGR5652BGR) { doTest(2, 3, CVTCODE(BGR5652BGR)); }
+OCL_TEST_P(CvtColor8u, BGR5652RGB) { doTest(2, 3, CVTCODE(BGR5652RGB)); }
+OCL_TEST_P(CvtColor8u, BGR5652BGRA) { doTest(2, 4, CVTCODE(BGR5652BGRA)); }
+OCL_TEST_P(CvtColor8u, BGR5652RGBA) { doTest(2, 4, CVTCODE(BGR5652RGBA)); }
 
-OCL_TEST_P(CvtColor, RGB2GRAY)
-{
-    doTest(3, 1, CVTCODE(RGB2GRAY));
-}
-OCL_TEST_P(CvtColor, GRAY2RGB)
-{
-    doTest(1, 3, CVTCODE(GRAY2RGB));
-};
+OCL_TEST_P(CvtColor8u, BGR5552BGR) { doTest(2, 3, CVTCODE(BGR5552BGR)); }
+OCL_TEST_P(CvtColor8u, BGR5552RGB) { doTest(2, 3, CVTCODE(BGR5552RGB)); }
+OCL_TEST_P(CvtColor8u, BGR5552BGRA) { doTest(2, 4, CVTCODE(BGR5552BGRA)); }
+OCL_TEST_P(CvtColor8u, BGR5552RGBA) { doTest(2, 4, CVTCODE(BGR5552RGBA)); }
+
+OCL_TEST_P(CvtColor8u, BGR2BGR565) { doTest(3, 2, CVTCODE(BGR2BGR565)); }
+OCL_TEST_P(CvtColor8u, RGB2BGR565) { doTest(3, 2, CVTCODE(RGB2BGR565)); }
+OCL_TEST_P(CvtColor8u, BGRA2BGR565) { doTest(4, 2, CVTCODE(BGRA2BGR565)); }
+OCL_TEST_P(CvtColor8u, RGBA2BGR565) { doTest(4, 2, CVTCODE(RGBA2BGR565)); }
 
-OCL_TEST_P(CvtColor, BGR2GRAY)
-{
-    doTest(3, 1, CVTCODE(BGR2GRAY));
-}
-OCL_TEST_P(CvtColor, GRAY2BGR)
-{
-    doTest(1, 3, CVTCODE(GRAY2BGR));
-};
+OCL_TEST_P(CvtColor8u, BGR2BGR555) { doTest(3, 2, CVTCODE(BGR2BGR555)); }
+OCL_TEST_P(CvtColor8u, RGB2BGR555) { doTest(3, 2, CVTCODE(RGB2BGR555)); }
+OCL_TEST_P(CvtColor8u, BGRA2BGR555) { doTest(4, 2, CVTCODE(BGRA2BGR555)); }
+OCL_TEST_P(CvtColor8u, RGBA2BGR555) { doTest(4, 2, CVTCODE(RGBA2BGR555)); }
 
-OCL_TEST_P(CvtColor, RGBA2GRAY)
-{
-    doTest(3, 1, CVTCODE(RGBA2GRAY));
-}
-OCL_TEST_P(CvtColor, GRAY2RGBA)
-{
-    doTest(1, 3, CVTCODE(GRAY2RGBA));
-};
+// RGB5x5 <-> Gray
 
-OCL_TEST_P(CvtColor, BGRA2GRAY)
-{
-    doTest(3, 1, CVTCODE(BGRA2GRAY));
-}
-OCL_TEST_P(CvtColor, GRAY2BGRA)
-{
-    doTest(1, 3, CVTCODE(GRAY2BGRA));
-};
+OCL_TEST_P(CvtColor8u, BGR5652GRAY) { doTest(2, 1, CVTCODE(BGR5652GRAY)); }
+OCL_TEST_P(CvtColor8u, BGR5552GRAY) { doTest(2, 1, CVTCODE(BGR5552GRAY)); }
 
-OCL_TEST_P(CvtColor, RGB2YUV)
-{
-    doTest(3, 3, CVTCODE(RGB2YUV));
-}
-OCL_TEST_P(CvtColor, BGR2YUV)
-{
-    doTest(3, 3, CVTCODE(BGR2YUV));
-}
-OCL_TEST_P(CvtColor, YUV2RGB)
-{
-    doTest(3, 3, CVTCODE(YUV2RGB));
-}
-OCL_TEST_P(CvtColor, YUV2BGR)
-{
-    doTest(3, 3, CVTCODE(YUV2BGR));
-}
-OCL_TEST_P(CvtColor, RGB2YCrCb)
-{
-    doTest(3, 3, CVTCODE(RGB2YCrCb));
-}
-OCL_TEST_P(CvtColor, BGR2YCrCb)
-{
-    doTest(3, 3, CVTCODE(BGR2YCrCb));
-}
+OCL_TEST_P(CvtColor8u, GRAY2BGR565) { doTest(1, 2, CVTCODE(GRAY2BGR565)); }
+OCL_TEST_P(CvtColor8u, GRAY2BGR555) { doTest(1, 2, CVTCODE(GRAY2BGR555)); }
 
-struct CvtColor_YUV420 : CvtColor
+// RGBA <-> mRGBA
+
+OCL_TEST_P(CvtColor8u, RGBA2mRGBA) { doTest(4, 4, CVTCODE(RGBA2mRGBA)); }
+OCL_TEST_P(CvtColor8u, mRGBA2RGBA) { doTest(4, 4, CVTCODE(mRGBA2RGBA)); }
+
+// YUV -> RGBA_NV12
+
+struct CvtColor_YUV420 :
+        public CvtColor
 {
     void random_roi(int channelsIn, int channelsOut)
     {
@@ -190,50 +278,36 @@ struct CvtColor_YUV420 : CvtColor
         roiSize.width *= 2;
         roiSize.height *= 3;
         Border srcBorder = randomBorder(0, use_roi ? MAX_VALUE : 0);
-        randomSubMat(src1, src1_roi, roiSize, srcBorder, srcType, 2, 100);
+        randomSubMat(src, src_roi, roiSize, srcBorder, srcType, 2, 100);
 
-        Border dst1Border = randomBorder(0, use_roi ? MAX_VALUE : 0);
-        randomSubMat(dst1, dst1_roi, roiSize, dst1Border, dstType, 5, 16);
+        Border dstBorder = randomBorder(0, use_roi ? MAX_VALUE : 0);
+        randomSubMat(dst, dst_roi, roiSize, dstBorder, dstType, 5, 16);
 
-        generateOclMat(gsrc1_whole, gsrc1_roi, src1, roiSize, srcBorder);
-        generateOclMat(gdst1_whole, gdst1_roi, dst1, roiSize, dst1Border);
+        generateOclMat(gsrc_whole, gsrc_roi, src, roiSize, srcBorder);
+        generateOclMat(gdst_whole, gdst_roi, dst, roiSize, dstBorder);
     }
 };
 
-OCL_TEST_P(CvtColor_YUV420, YUV2RGBA_NV12)
-{
-    doTest(1, 4, CV_YUV2RGBA_NV12);
-};
-
-OCL_TEST_P(CvtColor_YUV420, YUV2BGRA_NV12)
-{
-    doTest(1, 4, CV_YUV2BGRA_NV12);
-};
+OCL_TEST_P(CvtColor_YUV420, YUV2RGBA_NV12) { doTest(1, 4, CV_YUV2RGBA_NV12); }
+OCL_TEST_P(CvtColor_YUV420, YUV2BGRA_NV12) { doTest(1, 4, CV_YUV2BGRA_NV12); }
+OCL_TEST_P(CvtColor_YUV420, YUV2RGB_NV12) { doTest(1, 3, CV_YUV2RGB_NV12); }
+OCL_TEST_P(CvtColor_YUV420, YUV2BGR_NV12) { doTest(1, 3, CV_YUV2BGR_NV12); }
 
-OCL_TEST_P(CvtColor_YUV420, YUV2RGB_NV12)
-{
-    doTest(1, 3, CV_YUV2RGB_NV12);
-};
 
-OCL_TEST_P(CvtColor_YUV420, YUV2BGR_NV12)
-{
-    doTest(1, 3, CV_YUV2BGR_NV12);
-};
+INSTANTIATE_TEST_CASE_P(OCL_ImgProc, CvtColor8u,
+                            testing::Combine(testing::Values(MatDepth(CV_8U)), Bool()));
 
+INSTANTIATE_TEST_CASE_P(OCL_ImgProc, CvtColor8u32f,
+                            testing::Combine(testing::Values(MatDepth(CV_8U), MatDepth(CV_32F)), Bool()));
 
 INSTANTIATE_TEST_CASE_P(OCL_ImgProc, CvtColor,
                             testing::Combine(
                                 testing::Values(MatDepth(CV_8U), MatDepth(CV_16U), MatDepth(CV_32F)),
-                                Bool()
-                            )
-                        );
+                                Bool()));
 
 INSTANTIATE_TEST_CASE_P(OCL_ImgProc, CvtColor_YUV420,
                             testing::Combine(
                                 testing::Values(MatDepth(CV_8U)),
-                                Bool()
-                            )
-                        );
+                                Bool()));
 
-}
 #endif

diff --git a/modules/ocl/test/test_imgproc.cpp b/modules/ocl/test/test_imgproc.cpp
index d30def8..353beaa 100644 (file)
--- a/modules/ocl/test/test_imgproc.cpp
+++ b/modules/ocl/test/test_imgproc.cpp
@@ -275,23 +275,33 @@ OCL_TEST_P(CornerHarris, Mat)
 
 //////////////////////////////////integral/////////////////////////////////////////////////
 
-typedef ImgprocTestBase Integral;
+struct Integral :
+        public ImgprocTestBase
+{
+    int sdepth;
 
+    virtual void SetUp()
+    {
+        type = GET_PARAM(0);
+        blockSize = GET_PARAM(1);
+        sdepth = GET_PARAM(2);
+        useRoi = GET_PARAM(3);
+    }
+};
 OCL_TEST_P(Integral, Mat1)
 {
     for (int j = 0; j < LOOP_TIMES; j++)
     {
         random_roi();
 
-        ocl::integral(gsrc_roi, gdst_roi);
-        integral(src_roi, dst_roi);
+        ocl::integral(gsrc_roi, gdst_roi, sdepth);
+        integral(src_roi, dst_roi, sdepth);
 
         Near();
     }
 }
 
-// TODO wrong output type
-OCL_TEST_P(Integral, DISABLED_Mat2)
+OCL_TEST_P(Integral, Mat2)
 {
     Mat dst1;
     ocl::oclMat gdst1;
@@ -300,10 +310,12 @@ OCL_TEST_P(Integral, DISABLED_Mat2)
     {
         random_roi();
 
-        integral(src_roi, dst1, dst_roi);
-        ocl::integral(gsrc_roi, gdst1, gdst_roi);
+        integral(src_roi, dst_roi, dst1, sdepth);
+        ocl::integral(gsrc_roi, gdst_roi, gdst1, sdepth);
 
         Near();
+        if(gdst1.clCxt->supportsFeature(ocl::FEATURE_CL_DOUBLE))
+            EXPECT_MAT_NEAR(dst1, Mat(gdst1), 0.);
     }
 }
 
@@ -543,7 +555,7 @@ INSTANTIATE_TEST_CASE_P(Imgproc, CornerHarris, Combine(
 INSTANTIATE_TEST_CASE_P(Imgproc, Integral, Combine(
                             Values((MatType)CV_8UC1), // TODO does not work with CV_32F, CV_64F
                             Values(0), // not used
-                            Values(0), // not used
+                            Values((MatType)CV_32SC1, (MatType)CV_32FC1),
                             Bool()));
 
 INSTANTIATE_TEST_CASE_P(Imgproc, Threshold, Combine(

diff --git a/modules/ocl/test/test_kmeans.cpp b/modules/ocl/test/test_kmeans.cpp
index 6539c51..e1d0a17 100644 (file)
--- a/modules/ocl/test/test_kmeans.cpp
+++ b/modules/ocl/test/test_kmeans.cpp
@@ -61,7 +61,7 @@ PARAM_TEST_CASE(Kmeans, int, int, int)
     int type;
     int K;
     int flags;
-    cv::Mat src ;
+    Mat src ;
     ocl::oclMat d_src, d_dists;
 
     Mat labels, centers;
@@ -73,7 +73,7 @@ PARAM_TEST_CASE(Kmeans, int, int, int)
         flags = GET_PARAM(2);
 
         // MWIDTH=256, MHEIGHT=256. defined in utility.hpp
-        cv::Size size = cv::Size(MWIDTH, MHEIGHT);
+        Size size = Size(MWIDTH, MHEIGHT);
         src.create(size, type);
         int row_idx = 0;
         const int max_neighbour = MHEIGHT / K - 1;
@@ -159,15 +159,15 @@ INSTANTIATE_TEST_CASE_P(OCL_ML, Kmeans, Combine(
 
 /////////////////////////////// DistanceToCenters //////////////////////////////////////////
 
-CV_ENUM(DistType, NORM_L1, NORM_L2SQR);
+CV_ENUM(DistType, NORM_L1, NORM_L2SQR)
 
 PARAM_TEST_CASE(distanceToCenters, DistType, bool)
 {
-    cv::Size size;
     int distType;
     bool useRoi;
-    cv::Mat src, centers, src_roi, centers_roi;
-    cv::ocl::oclMat ocl_src, ocl_centers, ocl_src_roi, ocl_centers_roi;
+
+    Mat src, centers, src_roi, centers_roi;
+    ocl::oclMat ocl_src, ocl_centers, ocl_src_roi, ocl_centers_roi;
 
     virtual void SetUp()
     {
@@ -177,70 +177,59 @@ PARAM_TEST_CASE(distanceToCenters, DistType, bool)
 
     void random_roi()
     {
-        Size roiSize_src = randomSize(10,1000);
-        Size roiSize_centers = randomSize(10, 1000);
-        roiSize_src.width = roiSize_centers.width;
+        Size roiSizeSrc = randomSize(1, MAX_VALUE);
+        Size roiSizeCenters = randomSize(1, MAX_VALUE);
+        roiSizeSrc.width = roiSizeCenters.width;
 
-        Border srcBorder = randomBorder(0, useRoi ? 500 : 0);
-        randomSubMat(src, src_roi, roiSize_src, srcBorder, CV_32FC1, -SHRT_MAX, SHRT_MAX);
+        Border srcBorder = randomBorder(0, useRoi ? MAX_VALUE : 0);
+        randomSubMat(src, src_roi, roiSizeSrc, srcBorder, CV_32FC1, -MAX_VALUE, MAX_VALUE);
 
         Border centersBorder = randomBorder(0, useRoi ? 500 : 0);
-        randomSubMat(centers, centers_roi, roiSize_centers, centersBorder, CV_32FC1, -SHRT_MAX, SHRT_MAX);
-
-        for(int i = 0; i<centers.rows; i++)
-            centers.at<float>(i, randomInt(0,centers.cols-1)) = (float)randomDouble(SHRT_MAX, INT_MAX);
+        randomSubMat(centers, centers_roi, roiSizeCenters, centersBorder, CV_32FC1, -MAX_VALUE, MAX_VALUE);
 
-        generateOclMat(ocl_src, ocl_src_roi, src, roiSize_src, srcBorder);
-        generateOclMat(ocl_centers, ocl_centers_roi, centers, roiSize_centers, centersBorder);
+        for (int i = 0; i < centers.rows; i++)
+            centers.at<float>(i, randomInt(0, centers.cols)) = (float)randomDouble(SHRT_MAX, INT_MAX);
 
+        generateOclMat(ocl_src, ocl_src_roi, src, roiSizeSrc, srcBorder);
+        generateOclMat(ocl_centers, ocl_centers_roi, centers, roiSizeCenters, centersBorder);
     }
-
 };
 
 OCL_TEST_P(distanceToCenters, Accuracy)
 {
-    for(int j = 0; j< LOOP_TIMES; j++)
+    for (int j = 0; j < LOOP_TIMES; j++)
     {
         random_roi();
 
-        cv::ocl::oclMat ocl_dists;
-        cv::ocl::oclMat ocl_labels;
-
-        cv::ocl::distanceToCenters(ocl_dists,ocl_labels,ocl_src_roi, ocl_centers_roi, distType);
-
         Mat labels, dists;
-        ocl_labels.download(labels);
-        ocl_dists.download(dists);
+        ocl::distanceToCenters(ocl_src_roi, ocl_centers_roi, dists, labels, distType);
 
-        ASSERT_EQ(ocl_dists.cols, ocl_labels.rows);
+        EXPECT_EQ(dists.size(), labels.size());
 
         Mat batch_dists;
-
         cv::batchDistance(src_roi, centers_roi, batch_dists, CV_32FC1, noArray(), distType);
 
-        std::vector<double> gold_dists_v;
+        std::vector<float> gold_dists_v;
+        gold_dists_v.reserve(batch_dists.rows);
 
-        for(int i = 0; i<batch_dists.rows; i++)
+        for (int i = 0; i < batch_dists.rows; i++)
         {
             Mat r = batch_dists.row(i);
             double mVal;
             Point mLoc;
             minMaxLoc(r, &mVal, NULL, &mLoc, NULL);
 
-            int ocl_label = *(int*)labels.row(i).col(0).data;
-            ASSERT_EQ(mLoc.x, ocl_label);
+            int ocl_label = labels.at<int>(i, 0);
+            EXPECT_EQ(mLoc.x, ocl_label);
 
-            gold_dists_v.push_back(mVal);
+            gold_dists_v.push_back(static_cast<float>(mVal));
         }
-        Mat gold_dists(gold_dists_v);
-        dists.convertTo(dists, CV_64FC1);
-        double relative_error = cv::norm(gold_dists.t(), dists, NORM_INF|NORM_RELATIVE);
+
+        double relative_error = cv::norm(Mat(gold_dists_v), dists, NORM_INF | NORM_RELATIVE);
         ASSERT_LE(relative_error, 1e-5);
     }
 }
 
-
-INSTANTIATE_TEST_CASE_P (OCL_ML, distanceToCenters, Combine(DistType::all(), Bool()) );
-
+INSTANTIATE_TEST_CASE_P (OCL_ML, distanceToCenters, Combine(DistType::all(), Bool()));
 
 #endif

diff --git a/modules/ocl/test/test_warp.cpp b/modules/ocl/test/test_warp.cpp
index 05554ce..85f3375 100644 (file)
--- a/modules/ocl/test/test_warp.cpp
+++ b/modules/ocl/test/test_warp.cpp
@@ -156,6 +156,114 @@ OCL_TEST_P(WarpPerspective, Mat)
     }
 }
 
+// buildWarpPerspectiveMaps
+
+PARAM_TEST_CASE(BuildWarpPerspectiveMaps, bool, bool)
+{
+    bool useRoi, mapInverse;
+    Size dsize;
+
+    Mat xmap_whole, ymap_whole, xmap_roi, ymap_roi;
+    ocl::oclMat gxmap_whole, gymap_whole, gxmap_roi, gymap_roi;
+
+    void SetUp()
+    {
+        mapInverse = GET_PARAM(0);
+        useRoi = GET_PARAM(1);
+    }
+
+    void random_roi()
+    {
+        dsize = randomSize(1, MAX_VALUE);
+
+        Border xmapBorder = randomBorder(0, useRoi ? MAX_VALUE : 0);
+        randomSubMat(xmap_whole, xmap_roi, dsize, xmapBorder, CV_32FC1, -MAX_VALUE, MAX_VALUE);
+
+        Border ymapBorder = randomBorder(0, useRoi ? MAX_VALUE : 0);
+        randomSubMat(ymap_whole, ymap_roi, dsize, ymapBorder, CV_32FC1, -MAX_VALUE, MAX_VALUE);
+
+        generateOclMat(gxmap_whole, gxmap_roi, xmap_whole, dsize, xmapBorder);
+        generateOclMat(gymap_whole, gymap_roi, ymap_whole, dsize, ymapBorder);
+    }
+
+    void Near(double threshold = 0.0)
+    {
+        Mat whole, roi;
+        gxmap_whole.download(whole);
+        gxmap_roi.download(roi);
+
+        EXPECT_MAT_NEAR(xmap_whole, whole, threshold);
+        EXPECT_MAT_NEAR(xmap_roi, roi, threshold);
+    }
+
+    void Near1(double threshold = 0.0)
+    {
+        Mat whole, roi;
+        gymap_whole.download(whole);
+        gymap_roi.download(roi);
+
+        EXPECT_MAT_NEAR(ymap_whole, whole, threshold);
+        EXPECT_MAT_NEAR(ymap_roi, roi, threshold);
+    }
+};
+
+static void buildWarpPerspectiveMaps(const Mat &M, bool inverse, Size dsize, Mat &xmap, Mat &ymap)
+{
+    CV_Assert(M.rows == 3 && M.cols == 3);
+    CV_Assert(dsize.area() > 0);
+
+    xmap.create(dsize, CV_32FC1);
+    ymap.create(dsize, CV_32FC1);
+
+    float coeffs[3 * 3];
+    Mat coeffsMat(3, 3, CV_32F, (void *)coeffs);
+
+    if (inverse)
+        M.convertTo(coeffsMat, coeffsMat.type());
+    else
+    {
+        cv::Mat iM;
+        invert(M, iM);
+        iM.convertTo(coeffsMat, coeffsMat.type());
+    }
+
+    for (int y = 0; y < dsize.height; ++y)
+    {
+        float * const xmap_ptr = xmap.ptr<float>(y);
+        float * const ymap_ptr = ymap.ptr<float>(y);
+
+        for (int x = 0; x < dsize.width; ++x)
+        {
+            float coeff = 1.0f / (x * coeffs[6] + y * coeffs[7] + coeffs[8]);
+            xmap_ptr[x] = (x * coeffs[0] + y * coeffs[1] + coeffs[2]) * coeff;
+            ymap_ptr[x] = (x * coeffs[3] + y * coeffs[4] + coeffs[5]) * coeff;
+        }
+    }
+}
+
+OCL_TEST_P(BuildWarpPerspectiveMaps, Mat)
+{
+    for (int j = 0; j < LOOP_TIMES; j++)
+    {
+        random_roi();
+
+        float cols = static_cast<float>(MAX_VALUE), rows = static_cast<float>(MAX_VALUE);
+        float cols2 = cols / 2.0f, rows2 = rows / 2.0f;
+        Point2f sp[] = { Point2f(0.0f, 0.0f), Point2f(cols, 0.0f), Point2f(0.0f, rows), Point2f(cols, rows) };
+        Point2f dp[] = { Point2f(rng.uniform(0.0f, cols2), rng.uniform(0.0f, rows2)),
+            Point2f(rng.uniform(cols2, cols), rng.uniform(0.0f, rows2)),
+            Point2f(rng.uniform(0.0f, cols2), rng.uniform(rows2, rows)),
+            Point2f(rng.uniform(cols2, cols), rng.uniform(rows2, rows)) };
+        Mat M = getPerspectiveTransform(sp, dp);
+
+        buildWarpPerspectiveMaps(M, mapInverse, dsize, xmap_roi, ymap_roi);
+        ocl::buildWarpPerspectiveMaps(M, mapInverse, dsize, gxmap_roi, gymap_roi);
+
+        Near(5e-3);
+        Near1(5e-3);
+    }
+}
+
 /////////////////////////////////////////////////////////////////////////////////////////////////
 // remap
 
@@ -205,7 +313,12 @@ PARAM_TEST_CASE(Remap, MatDepth, Channels, pair<MatType, MatType>, Border, bool)
 
         Border map2Border = randomBorder(0, useRoi ? MAX_VALUE : 0);
         if (map2Type != noType)
-            randomSubMat(map2, map2_roi, dstROISize, map2Border, map2Type, -mapMaxValue, mapMaxValue);
+        {
+            int mapMinValue = -mapMaxValue;
+            if (map2Type == CV_16UC1 || map2Type == CV_16SC1)
+                mapMinValue = 0, mapMaxValue = INTER_TAB_SIZE2;
+            randomSubMat(map2, map2_roi, dstROISize, map2Border, map2Type, mapMinValue, mapMaxValue);
+        }
 
         generateOclMat(gsrc, gsrc_roi, src, srcROISize, srcBorder);
         generateOclMat(gdst, gdst_roi, dst, dstROISize, dstBorder);
@@ -285,10 +398,7 @@ PARAM_TEST_CASE(Resize, MatType, double, double, Interpolation, bool)
         dstRoiSize.height = cvRound(srcRoiSize.height * fy);
 
         if (dstRoiSize.area() == 0)
-        {
-            random_roi();
-            return;
-        }
+            return random_roi();
 
         Border srcBorder = randomBorder(0, useRoi ? MAX_VALUE : 0);
         randomSubMat(src, src_roi, srcRoiSize, srcBorder, type, -MAX_VALUE, MAX_VALUE);
@@ -338,10 +448,13 @@ INSTANTIATE_TEST_CASE_P(ImgprocWarp, WarpPerspective, Combine(
                             Bool(),
                             Bool()));
 
+INSTANTIATE_TEST_CASE_P(ImgprocWarp, BuildWarpPerspectiveMaps, Combine(Bool(), Bool()));
+
 INSTANTIATE_TEST_CASE_P(ImgprocWarp, Remap_INTER_LINEAR, Combine(
                             Values(CV_8U, CV_16U, CV_16S, CV_32F, CV_64F),
                             Values(1, 2, 3, 4),
                             Values(pair<MatType, MatType>((MatType)CV_32FC1, (MatType)CV_32FC1),
+                                   pair<MatType, MatType>((MatType)CV_16SC2, (MatType)CV_16UC1),
                                    pair<MatType, MatType>((MatType)CV_32FC2, noType)),
                             Values((Border)BORDER_CONSTANT,
                                    (Border)BORDER_REPLICATE,
@@ -355,6 +468,7 @@ INSTANTIATE_TEST_CASE_P(ImgprocWarp, Remap_INTER_NEAREST, Combine(
                             Values(1, 2, 3, 4),
                             Values(pair<MatType, MatType>((MatType)CV_32FC1, (MatType)CV_32FC1),
                                    pair<MatType, MatType>((MatType)CV_32FC2, noType),
+                                   pair<MatType, MatType>((MatType)CV_16SC2, (MatType)CV_16UC1),
                                    pair<MatType, MatType>((MatType)CV_16SC2, noType)),
                             Values((Border)BORDER_CONSTANT,
                                    (Border)BORDER_REPLICATE,
@@ -363,11 +477,18 @@ INSTANTIATE_TEST_CASE_P(ImgprocWarp, Remap_INTER_NEAREST, Combine(
                                    (Border)BORDER_REFLECT_101),
                             Bool()));
 
-INSTANTIATE_TEST_CASE_P(ImgprocWarp, Resize, Combine(
-                            Values(CV_8UC1, CV_8UC3, CV_8UC4, CV_32FC1, CV_32FC3, CV_32FC4),
-                            Values(0.5, 1.5, 2.0),
-                            Values(0.5, 1.5, 2.0),
+INSTANTIATE_TEST_CASE_P(ImgprocWarpResize, Resize, Combine(
+                            Values((MatType)CV_8UC1, CV_8UC3, CV_8UC4, CV_32FC1, CV_32FC3, CV_32FC4),
+                            Values(0.7, 0.4, 2.0),
+                            Values(0.3, 0.6, 2.0),
                             Values((Interpolation)INTER_NEAREST, (Interpolation)INTER_LINEAR),
                             Bool()));
 
+INSTANTIATE_TEST_CASE_P(ImgprocWarpResizeArea, Resize, Combine(
+                            Values((MatType)CV_8UC1, CV_8UC3, CV_8UC4, CV_32FC1, CV_32FC3, CV_32FC4),
+                            Values(0.7, 0.4, 0.5),
+                            Values(0.3, 0.6, 0.5),
+                            Values((Interpolation)INTER_AREA),
+                            Bool()));
+
 #endif // HAVE_OPENCL

diff --git a/modules/ocl/test/utility.cpp b/modules/ocl/test/utility.cpp
index f986042..c9cb0d8 100644 (file)
--- a/modules/ocl/test/utility.cpp
+++ b/modules/ocl/test/utility.cpp
@@ -231,7 +231,7 @@ double checkRectSimilarity(Size sz, std::vector<Rect>& ob1, std::vector<Rect>& o
     return final_test_result;
 }
 
-void showDiff(const Mat& gold, const Mat& actual, double eps, bool alwaysShow)
+void showDiff(const Mat& src, const Mat& gold, const Mat& actual, double eps, bool alwaysShow)
 {
     Mat diff, diff_thresh;
     absdiff(gold, actual, diff);
@@ -240,10 +240,18 @@ void showDiff(const Mat& gold, const Mat& actual, double eps, bool alwaysShow)
 
     if (alwaysShow || cv::countNonZero(diff_thresh.reshape(1)) > 0)
     {
+#if 0
+        std::cout << "Src: " << std::endl << src << std::endl;
+        std::cout << "Reference: " << std::endl << gold << std::endl;
+        std::cout << "OpenCL: " << std::endl << actual << std::endl;
+#endif
+
+        namedWindow("src", WINDOW_NORMAL);
         namedWindow("gold", WINDOW_NORMAL);
         namedWindow("actual", WINDOW_NORMAL);
         namedWindow("diff", WINDOW_NORMAL);
 
+        imshow("src", src);
         imshow("gold", gold);
         imshow("actual", actual);
         imshow("diff", diff);

diff --git a/modules/ocl/test/utility.hpp b/modules/ocl/test/utility.hpp
index d7ae1b9..a1fe3ff 100644 (file)
--- a/modules/ocl/test/utility.hpp
+++ b/modules/ocl/test/utility.hpp
@@ -52,7 +52,7 @@ extern int LOOP_TIMES;
 
 namespace cvtest {
 
-void showDiff(const Mat& gold, const Mat& actual, double eps, bool alwaysShow = false);
+void showDiff(const Mat& src, const Mat& gold, const Mat& actual, double eps, bool alwaysShow = false);
 
 cv::ocl::oclMat createMat_ocl(cv::RNG& rng, Size size, int type, bool useRoi);
 cv::ocl::oclMat loadMat_ocl(cv::RNG& rng, const Mat& m, bool useRoi);
@@ -262,7 +262,7 @@ CV_ENUM(NormCode, NORM_INF, NORM_L1, NORM_L2, NORM_TYPE_MASK, NORM_RELATIVE, NOR
 CV_ENUM(ReduceOp, CV_REDUCE_SUM, CV_REDUCE_AVG, CV_REDUCE_MAX, CV_REDUCE_MIN)
 CV_ENUM(MorphOp, MORPH_OPEN, MORPH_CLOSE, MORPH_GRADIENT, MORPH_TOPHAT, MORPH_BLACKHAT)
 CV_ENUM(ThreshOp, THRESH_BINARY, THRESH_BINARY_INV, THRESH_TRUNC, THRESH_TOZERO, THRESH_TOZERO_INV)
-CV_ENUM(Interpolation, INTER_NEAREST, INTER_LINEAR, INTER_CUBIC)
+CV_ENUM(Interpolation, INTER_NEAREST, INTER_LINEAR, INTER_CUBIC, INTER_AREA)
 CV_ENUM(Border, BORDER_REFLECT101, BORDER_REPLICATE, BORDER_CONSTANT, BORDER_REFLECT, BORDER_WRAP)
 CV_ENUM(TemplateMethod, TM_SQDIFF, TM_SQDIFF_NORMED, TM_CCORR, TM_CCORR_NORMED, TM_CCOEFF, TM_CCOEFF_NORMED)
 

diff --git a/modules/ts/include/opencv2/ts/ts_perf.hpp b/modules/ts/include/opencv2/ts/ts_perf.hpp
index 6a0e921..9238b3e 100644 (file)
--- a/modules/ts/include/opencv2/ts/ts_perf.hpp
+++ b/modules/ts/include/opencv2/ts/ts_perf.hpp
@@ -243,6 +243,7 @@ typedef struct CV_EXPORTS performance_metrics
         TERM_TIME = 1,
         TERM_INTERRUPT = 2,
         TERM_EXCEPTION = 3,
+        TERM_SKIP_TEST = 4, // there are some limitations and test should be skipped
         TERM_UNKNOWN = -1
     };
 
@@ -279,6 +280,8 @@ public:
     static enum PERF_STRATEGY getPerformanceStrategy();
     static enum PERF_STRATEGY setPerformanceStrategy(enum PERF_STRATEGY strategy);
 
+    class PerfSkipTestException: public cv::Exception {};
+
 protected:
     virtual void PerfTestBody() = 0;
 

diff --git a/modules/ts/src/ts_perf.cpp b/modules/ts/src/ts_perf.cpp
index 319076c..3bcb5c1 100644 (file)
--- a/modules/ts/src/ts_perf.cpp
+++ b/modules/ts/src/ts_perf.cpp
@@ -1175,7 +1175,14 @@ void TestBase::reportMetrics(bool toJUnitXML)
 {
     performance_metrics& m = calcMetrics();
 
-    if (toJUnitXML)
+    if (m.terminationReason == performance_metrics::TERM_SKIP_TEST)
+    {
+        if (toJUnitXML)
+        {
+            RecordProperty("custom_status", "skipped");
+        }
+    }
+    else if (toJUnitXML)
     {
         RecordProperty("bytesIn", (int)m.bytesIn);
         RecordProperty("bytesOut", (int)m.bytesOut);
@@ -1267,21 +1274,30 @@ void TestBase::SetUp()
 
 void TestBase::TearDown()
 {
-    if (!HasFailure() && !verified)
-        ADD_FAILURE() << "The test has no sanity checks. There should be at least one check at the end of performance test.";
-
-    validateMetrics();
-    if (HasFailure())
-        reportMetrics(false);
+    if (metrics.terminationReason == performance_metrics::TERM_SKIP_TEST)
+    {
+        LOGI("\tTest was skipped");
+        GTEST_SUCCEED() << "Test was skipped";
+    }
     else
     {
-        const ::testing::TestInfo* const test_info = ::testing::UnitTest::GetInstance()->current_test_info();
-        const char* type_param = test_info->type_param();
-        const char* value_param = test_info->value_param();
-        if (value_param) printf("[ VALUE    ] \t%s\n", value_param), fflush(stdout);
-        if (type_param)  printf("[ TYPE     ] \t%s\n", type_param), fflush(stdout);
-        reportMetrics(true);
+        if (!HasFailure() && !verified)
+            ADD_FAILURE() << "The test has no sanity checks. There should be at least one check at the end of performance test.";
+
+        validateMetrics();
+        if (HasFailure())
+        {
+            reportMetrics(false);
+            return;
+        }
     }
+
+    const ::testing::TestInfo* const test_info = ::testing::UnitTest::GetInstance()->current_test_info();
+    const char* type_param = test_info->type_param();
+    const char* value_param = test_info->value_param();
+    if (value_param) printf("[ VALUE    ] \t%s\n", value_param), fflush(stdout);
+    if (type_param)  printf("[ TYPE     ] \t%s\n", type_param), fflush(stdout);
+    reportMetrics(true);
 }
 
 std::string TestBase::getDataPath(const std::string& relativePath)
@@ -1331,6 +1347,11 @@ void TestBase::RunPerfTestBody()
     {
         this->PerfTestBody();
     }
+    catch(PerfSkipTestException&)
+    {
+        metrics.terminationReason = performance_metrics::TERM_SKIP_TEST;
+        return;
+    }
     catch(PerfEarlyExitException&)
     {
         metrics.terminationReason = performance_metrics::TERM_INTERRUPT;

diff --git a/modules/world/CMakeLists.txt b/modules/world/CMakeLists.txt
index f18635e..f65447d 100644 (file)
--- a/modules/world/CMakeLists.txt
+++ b/modules/world/CMakeLists.txt
@@ -66,7 +66,9 @@ foreach(m ${OPENCV_MODULE_${the_module}_DEPS})
           endif()
         endif()
         string(REPLACE ".." "__" srcname "${srcname}")
-        get_filename_component(srcname_we ${srcname} NAME_WE)
+        #NAME_WE intentionally not used since it interprets first period as start of extension (http://cmake.org/Bug/view.php?id=12282)
+        get_filename_component(srcname_we "${srcname}" NAME)
+        string(REGEX REPLACE "\\.[^.]+$" "" srcname_we "${srcname_we}")
         string(REGEX REPLACE <SRC_NAME_WE> "${srcname_we}" objpath2 "${objpath1}")
         string(REGEX REPLACE <RELATIVE_SRC_NAME> "${srcname}" objpath3 "${objpath2}")
       endif()
@@ -92,7 +94,9 @@ macro(ios_include_3party_libs)
       endif()
 
       string(REPLACE ".." "__" srcname "${srcname}")
-      get_filename_component(srcname_we ${srcname} NAME_WE)
+      #NAME_WE intentionally not used since it interprets first period as start of extension (http://cmake.org/Bug/view.php?id=12282)
+      get_filename_component(srcname_we "${srcname}" NAME)
+      string(REGEX REPLACE "\\.[^.]+$" "" srcname_we "${srcname_we}")
       string(REGEX REPLACE <SRC_NAME_WE> "${srcname_we}" objpath2 "${objpath1}")
       string(REGEX REPLACE <RELATIVE_SRC_NAME> "${srcname}" objpath3 "${objpath2}")
 

diff --git a/platforms/android/android.toolchain.cmake b/platforms/android/android.toolchain.cmake
index a9d4d7c..1b9880e 100644 (file)
--- a/platforms/android/android.toolchain.cmake
+++ b/platforms/android/android.toolchain.cmake
@@ -318,7 +318,7 @@ set( CMAKE_SYSTEM_VERSION 1 )
 # rpath makes low sence for Android
 set( CMAKE_SKIP_RPATH TRUE CACHE BOOL "If set, runtime paths are not added when using shared libraries." )
 
-set( ANDROID_SUPPORTED_NDK_VERSIONS ${ANDROID_EXTRA_NDK_VERSIONS} -r9 -r8e -r8d -r8c -r8b -r8 -r7c -r7b -r7 -r6b -r6 -r5c -r5b -r5 "" )
+set( ANDROID_SUPPORTED_NDK_VERSIONS ${ANDROID_EXTRA_NDK_VERSIONS} -r9b -r9 -r8e -r8d -r8c -r8b -r8 -r7c -r7b -r7 -r6b -r6 -r5c -r5b -r5 "" )
 if(NOT DEFINED ANDROID_NDK_SEARCH_PATHS)
  if( CMAKE_HOST_WIN32 )
   file( TO_CMAKE_PATH "$ENV{PROGRAMFILES}" ANDROID_NDK_SEARCH_PATHS )
@@ -634,27 +634,30 @@ endif()
 
 macro( __GLOB_NDK_TOOLCHAINS __availableToolchainsVar __availableToolchainsLst __toolchain_subpath )
  foreach( __toolchain ${${__availableToolchainsLst}} )
-  if( "${__toolchain}" MATCHES "-clang3[.][0-9]$" AND NOT EXISTS "${ANDROID_NDK_TOOLCHAINS_PATH}/${__toolchain}${__toolchain_subpath}" )
-   string( REGEX REPLACE "-clang3[.][0-9]$" "-4.6" __gcc_toolchain "${__toolchain}" )
-  else()
-   set( __gcc_toolchain "${__toolchain}" )
-  endif()
-  __DETECT_TOOLCHAIN_MACHINE_NAME( __machine "${ANDROID_NDK_TOOLCHAINS_PATH}/${__gcc_toolchain}${__toolchain_subpath}" )
-  if( __machine )
-   string( REGEX MATCH "[0-9]+[.][0-9]+([.][0-9x]+)?$" __version "${__gcc_toolchain}" )
-   if( __machine MATCHES i686 )
-    set( __arch "x86" )
-   elseif( __machine MATCHES arm )
-    set( __arch "arm" )
-   elseif( __machine MATCHES mipsel )
-    set( __arch "mipsel" )
+  # Skip renderscript folder. It's not C++ toolchain
+  if (NOT ${__toolchain} STREQUAL "renderscript")
+   if( "${__toolchain}" MATCHES "-clang3[.][0-9]$" AND NOT EXISTS "${ANDROID_NDK_TOOLCHAINS_PATH}/${__toolchain}${__toolchain_subpath}" )
+     string( REGEX REPLACE "-clang3[.][0-9]$" "-4.6" __gcc_toolchain "${__toolchain}" )
+   else()
+     set( __gcc_toolchain "${__toolchain}" )
+   endif()
+   __DETECT_TOOLCHAIN_MACHINE_NAME( __machine "${ANDROID_NDK_TOOLCHAINS_PATH}/${__gcc_toolchain}${__toolchain_subpath}" )
+   if( __machine )
+    string( REGEX MATCH "[0-9]+[.][0-9]+([.][0-9x]+)?$" __version "${__gcc_toolchain}" )
+    if( __machine MATCHES i686 )
+     set( __arch "x86" )
+    elseif( __machine MATCHES arm )
+     set( __arch "arm" )
+    elseif( __machine MATCHES mipsel )
+     set( __arch "mipsel" )
+    endif()
+    list( APPEND __availableToolchainMachines "${__machine}" )
+    list( APPEND __availableToolchainArchs "${__arch}" )
+    list( APPEND __availableToolchainCompilerVersions "${__version}" )
+    list( APPEND ${__availableToolchainsVar} "${__toolchain}" )
    endif()
-   list( APPEND __availableToolchainMachines "${__machine}" )
-   list( APPEND __availableToolchainArchs "${__arch}" )
-   list( APPEND __availableToolchainCompilerVersions "${__version}" )
-   list( APPEND ${__availableToolchainsVar} "${__toolchain}" )
+   unset( __gcc_toolchain )
   endif()
-  unset( __gcc_toolchain )
  endforeach()
 endmacro()
 

diff --git a/platforms/scripts/camera_build.conf b/platforms/scripts/camera_build.conf
index 7392f91..e41dbcf 100644 (file)
--- a/platforms/scripts/camera_build.conf
+++ b/platforms/scripts/camera_build.conf
@@ -1,27 +1,31 @@
 # make target;        arch; API level; Android Source Code Root
-native_camera_r2.2.0; armeabi;     8;  /home/alexander/Projects/AndroidSource/2.2.2
-native_camera_r2.2.0; armeabi-v7a; 8;  /home/alexander/Projects/AndroidSource/2.2.2
-native_camera_r2.3.3; armeabi;     9;  /home/alexander/Projects/AndroidSource/2.3.3
-native_camera_r2.3.3; armeabi-v7a; 9;  /home/alexander/Projects/AndroidSource/2.3.3
-native_camera_r2.3.3; x86;         9;  /home/alexander/Projects/AndroidSource/2.3.3
-native_camera_r3.0.1; armeabi;     9;  /home/alexander/Projects/AndroidSource/3.0.1
-native_camera_r3.0.1; armeabi-v7a; 9;  /home/alexander/Projects/AndroidSource/3.0.1
-native_camera_r3.0.1; x86;         9;  /home/alexander/Projects/AndroidSource/3.0.1
-native_camera_r4.0.3; armeabi;     14; /home/alexander/Projects/AndroidSource/4.0.3
-native_camera_r4.0.3; armeabi-v7a; 14; /home/alexander/Projects/AndroidSource/4.0.3
-native_camera_r4.0.3; x86;         14; /home/alexander/Projects/AndroidSource/4.0.3
-native_camera_r4.0.3; mips;        14; /home/alexander/Projects/AndroidSource/4.0.3_mips
-native_camera_r4.0.0; armeabi;     14; /home/alexander/Projects/AndroidSource/4.0.0
-native_camera_r4.0.0; armeabi-v7a; 14; /home/alexander/Projects/AndroidSource/4.0.0
-native_camera_r4.1.1; armeabi;     14; /home/alexander/Projects/AndroidSource/4.1.1
-native_camera_r4.1.1; armeabi-v7a; 14; /home/alexander/Projects/AndroidSource/4.1.1
-native_camera_r4.1.1; x86;         14; /home/alexander/Projects/AndroidSource/4.1.1
-native_camera_r4.1.1; mips;        14; /home/alexander/Projects/AndroidSource/4.1.1_mips
-native_camera_r4.2.0; armeabi-v7a; 14; /home/alexander/Projects/AndroidSource/4.2
-native_camera_r4.2.0; armeabi;     14; /home/alexander/Projects/AndroidSource/4.2
-native_camera_r4.2.0; x86;         14; /home/alexander/Projects/AndroidSource/4.2
-native_camera_r4.2.0; mips;        14; /home/alexander/Projects/AndroidSource/4.2
-native_camera_r4.3.0; armeabi;     14; /home/alexander/Projects/AndroidSource/4.3
-native_camera_r4.3.0; armeabi-v7a; 14; /home/alexander/Projects/AndroidSource/4.3
-native_camera_r4.3.0; x86;         14; /home/alexander/Projects/AndroidSource/4.3
-native_camera_r4.3.0; mips;        14; /home/alexander/Projects/AndroidSource/4.3
+native_camera_r2.2.0; armeabi;     8;  $ANDROID_STUB_ROOT/2.2.2
+native_camera_r2.2.0; armeabi-v7a; 8;  $ANDROID_STUB_ROOT/2.2.2
+native_camera_r2.3.3; armeabi;     9;  $ANDROID_STUB_ROOT/2.3.3
+native_camera_r2.3.3; armeabi-v7a; 9;  $ANDROID_STUB_ROOT/2.3.3
+native_camera_r2.3.3; x86;         9;  $ANDROID_STUB_ROOT/2.3.3
+native_camera_r3.0.1; armeabi;     9;  $ANDROID_STUB_ROOT/3.0.1
+native_camera_r3.0.1; armeabi-v7a; 9;  $ANDROID_STUB_ROOT/3.0.1
+native_camera_r3.0.1; x86;         9;  $ANDROID_STUB_ROOT/3.0.1
+native_camera_r4.0.3; armeabi;     14; $ANDROID_STUB_ROOT/4.0.3
+native_camera_r4.0.3; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.0.3
+native_camera_r4.0.3; x86;         14; $ANDROID_STUB_ROOT/4.0.3
+native_camera_r4.0.3; mips;        14; $ANDROID_STUB_ROOT/4.0.3_mips
+native_camera_r4.0.0; armeabi;     14; $ANDROID_STUB_ROOT/4.0.0
+native_camera_r4.0.0; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.0.0
+native_camera_r4.1.1; armeabi;     14; $ANDROID_STUB_ROOT/4.1.1
+native_camera_r4.1.1; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.1.1
+native_camera_r4.1.1; x86;         14; $ANDROID_STUB_ROOT/4.1.1
+native_camera_r4.1.1; mips;        14; $ANDROID_STUB_ROOT/4.1.1
+native_camera_r4.2.0; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.2.0
+native_camera_r4.2.0; armeabi;     14; $ANDROID_STUB_ROOT/4.2.0
+native_camera_r4.2.0; x86;         14; $ANDROID_STUB_ROOT/4.2.0
+native_camera_r4.2.0; mips;        14; $ANDROID_STUB_ROOT/4.2.0
+native_camera_r4.3.0; armeabi;     14; $ANDROID_STUB_ROOT/4.3.0
+native_camera_r4.3.0; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.3.0
+native_camera_r4.3.0; x86;         14; $ANDROID_STUB_ROOT/4.3.0
+native_camera_r4.3.0; mips;        14; $ANDROID_STUB_ROOT/4.3.0
+native_camera_r4.4.0; armeabi;     14; $ANDROID_STUB_ROOT/4.4.0
+native_camera_r4.4.0; armeabi-v7a; 14; $ANDROID_STUB_ROOT/4.4.0
+native_camera_r4.4.0; x86;         14; $ANDROID_STUB_ROOT/4.4.0
+native_camera_r4.4.0; mips;        14; $ANDROID_STUB_ROOT/4.4.0

diff --git a/platforms/scripts/cmake_android_all_cameras.py b/platforms/scripts/cmake_android_all_cameras.py
index c160df0..1ad69f4 100755 (executable)
--- a/platforms/scripts/cmake_android_all_cameras.py
+++ b/platforms/scripts/cmake_android_all_cameras.py
@@ -7,6 +7,12 @@ import shutil
 ScriptHome = os.path.split(sys.argv[0])[0]
 ConfFile = open(os.path.join(ScriptHome, "camera_build.conf"), "rt")
 HomeDir = os.getcwd()
+
+stub = os.environ.get("ANDROID_STUB_ROOT", "")
+
+if (stub == ""):
+    print("Warning: ANDROID_STUB_ROOT environment variable is not set or is empty")
+
 for s in ConfFile.readlines():
     s = s[0:s.find("#")]
     if (not s):
@@ -20,6 +26,7 @@ for s in ConfFile.readlines():
     NativeApiLevel = str.strip(keys[2])
     AndroidTreeRoot = str.strip(keys[3])
     AndroidTreeRoot = str.strip(AndroidTreeRoot, "\n")
+    AndroidTreeRoot = os.path.expandvars(AndroidTreeRoot)
     print("Building %s for %s" % (MakeTarget, Arch))
     BuildDir = os.path.join(HomeDir, MakeTarget + "_" + Arch)
 

diff --git a/samples/cpp/3calibration.cpp b/samples/cpp/3calibration.cpp
index 061e977..08ba397 100644 (file)
--- a/samples/cpp/3calibration.cpp
+++ b/samples/cpp/3calibration.cpp
@@ -271,7 +271,7 @@ int main( int argc, char** argv )
             {
                 vector<Point2f> ptvec;
                 imageSize = view.size();
-                cvtColor(view, viewGray, CV_BGR2GRAY);
+                cvtColor(view, viewGray, COLOR_BGR2GRAY);
                 bool found = findChessboardCorners( view, boardSize, ptvec, CV_CALIB_CB_ADAPTIVE_THRESH );
 
                 drawChessboardCorners( view, boardSize, Mat(ptvec), found );

diff --git a/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping.cpp b/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping.cpp
index e89003d..be3dcd3 100644 (file)
--- a/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping.cpp
+++ b/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping.cpp
@@ -1,10 +1,10 @@
 
 //============================================================================
-// Name        : HighDynamicRange_RetinaCompression.cpp
+// Name        : OpenEXRimages_HDR_Retina_toneMapping.cpp
 // Author      : Alexandre Benoit (benoit.alexandre.vision@gmail.com)
 // Version     : 0.1
 // Copyright   : Alexandre Benoit, LISTIC Lab, july 2011
-// Description : HighDynamicRange compression (tone mapping) with the help of the Gipsa/Listic's retina in C++, Ansi-style
+// Description : HighDynamicRange retina tone mapping with the help of the Gipsa/Listic's retina in C++, Ansi-style
 //============================================================================
 
 #include <iostream>
@@ -69,7 +69,7 @@ static void drawPlot(const cv::Mat curve, const std::string figureTitle, const i
      {
          cv::Mat rgbIntImg;
          outputMat.convertTo(rgbIntImg, CV_8UC3);
-         cvtColor(rgbIntImg, intGrayImage, CV_BGR2GRAY);
+         cvtColor(rgbIntImg, intGrayImage, cv::COLOR_BGR2GRAY);
      }
 
      // get histogram density probability in order to cut values under above edges limits (here 5-95%)... usefull for HDR pixel errors cancellation

diff --git a/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping_video.cpp b/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping_video.cpp
index fc3ac83..509f3e8 100644 (file)
--- a/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping_video.cpp
+++ b/samples/cpp/OpenEXRimages_HDR_Retina_toneMapping_video.cpp
@@ -4,7 +4,7 @@
 // Author      : Alexandre Benoit (benoit.alexandre.vision@gmail.com)
 // Version     : 0.2
 // Copyright   : Alexandre Benoit, LISTIC Lab, december 2011
-// Description : HighDynamicRange compression (tone mapping) for image sequences with the help of the Gipsa/Listic's retina in C++, Ansi-style
+// Description : HighDynamicRange retina tone mapping for image sequences with the help of the Gipsa/Listic's retina in C++, Ansi-style
 // Known issues: the input OpenEXR sequences can have bad computed pixels that should be removed
 //               => a simple method consists of cutting histogram edges (a slider for this on the UI is provided)
 //               => however, in image sequences, this histogramm cut must be done in an elegant way from frame to frame... still not done...
@@ -92,7 +92,7 @@ static void rescaleGrayLevelMat(const cv::Mat &inputMat, cv::Mat &outputMat, con
      {
          cv::Mat rgbIntImg;
          normalisedImage.convertTo(rgbIntImg, CV_8UC3);
-         cvtColor(rgbIntImg, intGrayImage, CV_BGR2GRAY);
+         cvtColor(rgbIntImg, intGrayImage, cv::COLOR_BGR2GRAY);
      }
 
      // get histogram density probability in order to cut values under above edges limits (here 5-95%)... usefull for HDR pixel errors cancellation

diff --git a/samples/cpp/Qt_sample/main.cpp b/samples/cpp/Qt_sample/main.cpp
index 105b973..8794539 100644 (file)
--- a/samples/cpp/Qt_sample/main.cpp
+++ b/samples/cpp/Qt_sample/main.cpp
@@ -4,7 +4,7 @@
 
 #include <iostream>
 #include <vector>
-#include <opencv/highgui.h>
+#include <opencv2/highgui/highgui.hpp>
 
 #if defined WIN32 || defined _WIN32 || defined WINCE
     #include <windows.h>
@@ -20,8 +20,7 @@
     #include <GL/gl.h>
 #endif
 
-#include <opencv/cxcore.h>
-#include <opencv/cv.h>
+#include <opencv2/core/core.hpp>
 
 using namespace std;
 using namespace cv;
@@ -111,12 +110,12 @@ static void initPOSIT(std::vector<CvPoint3D32f> *modelPoints)
     modelPoints->push_back(cvPoint3D32f(0.0f, CUBE_SIZE, 0.0f));
 }
 
-static void foundCorners(vector<CvPoint2D32f> *srcImagePoints,IplImage* source, IplImage* grayImage)
+static void foundCorners(vector<CvPoint2D32f> *srcImagePoints, const Mat& source, Mat& grayImage)
 {
-    cvCvtColor(source,grayImage,CV_RGB2GRAY);
-    cvSmooth( grayImage, grayImage,CV_GAUSSIAN,11);
-    cvNormalize(grayImage, grayImage, 0, 255, CV_MINMAX);
-    cvThreshold( grayImage, grayImage, 26, 255, CV_THRESH_BINARY_INV);//25
+    cvtColor(source, grayImage, COLOR_RGB2GRAY);
+    GaussianBlur(grayImage, grayImage, Size(11,11), 0, 0);
+    normalize(grayImage, grayImage, 0, 255, NORM_MINMAX);
+    threshold(grayImage, grayImage, 26, 255, THRESH_BINARY_INV); //25
 
     Mat MgrayImage = grayImage;
     //For debug
@@ -189,8 +188,8 @@ static void foundCorners(vector<CvPoint2D32f> *srcImagePoints,IplImage* source,
         for(size_t i = 0 ; i<srcImagePoints_temp.size(); i++ )
         {
             ss<<i;
-            circle(Msource,srcImagePoints->at(i),5,CV_RGB(255,0,0));
-            putText( Msource, ss.str(), srcImagePoints->at(i),CV_FONT_HERSHEY_SIMPLEX,1,CV_RGB(255,0,0));
+            circle(Msource,srcImagePoints->at(i),5,Scalar(0,0,255));
+            putText(Msource,ss.str(),srcImagePoints->at(i),FONT_HERSHEY_SIMPLEX,1,Scalar(0,0,255));
             ss.str("");
 
             //new coordinate system in the middle of the frame and reversed (camera coordinate system)
@@ -224,19 +223,19 @@ static void createOpenGLMatrixFrom(float *posePOSIT,const CvMatr32f &rotationMat
 int main(void)
 {
     help();
-    CvCapture* video = cvCaptureFromFile("cube4.avi");
-    CV_Assert(video);
+    VideoCapture video("cube4.avi");
+    CV_Assert(video.isOpened());
 
-    IplImage* source = cvCreateImage(cvGetSize(cvQueryFrame(video)),8,3);
-    IplImage* grayImage = cvCreateImage(cvGetSize(cvQueryFrame(video)),8,1);
+    Mat source, grayImage;
 
-    cvNamedWindow("original",CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);
-    cvNamedWindow("POSIT",CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);
+    video >> source;
+
+    namedWindow("original", WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);
+    namedWindow("POSIT", WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);
     displayOverlay("POSIT", "We lost the 4 corners' detection quite often (the red circles disappear). This demo is only to illustrate how to use OpenGL callback.\n -- Press ESC to exit.", 10000);
-    //For debug
-    //cvNamedWindow("tempGray",CV_WINDOW_AUTOSIZE);
+
     float OpenGLMatrix[]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
-    cvSetOpenGlDrawCallback("POSIT",on_opengl,OpenGLMatrix);
+    setOpenGlDrawCallback("POSIT",on_opengl,OpenGLMatrix);
 
     vector<CvPoint3D32f> modelPoints;
     initPOSIT(&modelPoints);
@@ -251,26 +250,22 @@ int main(void)
     vector<CvPoint2D32f> srcImagePoints(4,cvPoint2D32f(0,0));
 
 
-    while(cvWaitKey(33) != 27)
+    while(waitKey(33) != 27)
     {
-        source=cvQueryFrame(video);
-        cvShowImage("original",source);
+        video >> source;
+        imshow("original",source);
 
         foundCorners(&srcImagePoints,source,grayImage);
         cvPOSIT( positObject, &srcImagePoints[0], FOCAL_LENGTH, criteria, rotation_matrix, translation_vector );
         createOpenGLMatrixFrom(OpenGLMatrix,rotation_matrix,translation_vector);
 
-        cvShowImage("POSIT",source);
-        //For debug
-        //cvShowImage("tempGray",grayImage);
+        imshow("POSIT",source);
 
-        if (cvGetCaptureProperty(video,CV_CAP_PROP_POS_AVI_RATIO)>0.99)
-            cvSetCaptureProperty(video,CV_CAP_PROP_POS_AVI_RATIO,0);
+        if (VideoCapture::get(video,CV_CAP_PROP_POS_AVI_RATIO)>0.99)
+            VideoCapture::get(video,CV_CAP_PROP_POS_AVI_RATIO,0);
     }
 
-    cvDestroyAllWindows();
-    cvReleaseImage(&grayImage);
-    cvReleaseCapture(&video);
+    destroyAllWindows();
     cvReleasePOSITObject(&positObject);
 
     return 0;

diff --git a/samples/cpp/bagofwords_classification.cpp b/samples/cpp/bagofwords_classification.cpp
index e24a770..39342c4 100644 (file)
--- a/samples/cpp/bagofwords_classification.cpp
+++ b/samples/cpp/bagofwords_classification.cpp
@@ -54,10 +54,6 @@ static void help(char** argv)
     << "\n";
 }
 
-
-
-
-
 static void makeDir( const string& dir )
 {
 #if defined WIN32 || defined _WIN32

diff --git a/samples/cpp/bgfg_segm.cpp b/samples/cpp/bgfg_segm.cpp
index aff7d30..f0d313e 100644 (file)
--- a/samples/cpp/bgfg_segm.cpp
+++ b/samples/cpp/bgfg_segm.cpp
@@ -45,10 +45,10 @@ int main(int argc, const char** argv)
         return -1;
     }
 
-    namedWindow("image", CV_WINDOW_NORMAL);
-    namedWindow("foreground mask", CV_WINDOW_NORMAL);
-    namedWindow("foreground image", CV_WINDOW_NORMAL);
-    namedWindow("mean background image", CV_WINDOW_NORMAL);
+    namedWindow("image", WINDOW_NORMAL);
+    namedWindow("foreground mask", WINDOW_NORMAL);
+    namedWindow("foreground image", WINDOW_NORMAL);
+    namedWindow("mean background image", WINDOW_NORMAL);
 
     BackgroundSubtractorMOG2 bg_model;//(100, 3, 0.3, 5);
 

diff --git a/samples/cpp/build3dmodel.cpp b/samples/cpp/build3dmodel.cpp
index b896a39..6700fcc 100644 (file)
--- a/samples/cpp/build3dmodel.cpp
+++ b/samples/cpp/build3dmodel.cpp
@@ -418,7 +418,7 @@ static void build3dmodel( const Ptr<FeatureDetector>& detector,
     for( size_t i = 0; i < nimages; i++ )
     {
         Mat img = imread(imageList[i], 1), gray;
-        cvtColor(img, gray, CV_BGR2GRAY);
+        cvtColor(img, gray, COLOR_BGR2GRAY);
 
         vector<KeyPoint> keypoints;
         detector->detect(gray, keypoints);

diff --git a/samples/cpp/calibration.cpp b/samples/cpp/calibration.cpp
index f4ff810..2d20776 100644 (file)
--- a/samples/cpp/calibration.cpp
+++ b/samples/cpp/calibration.cpp
@@ -461,7 +461,7 @@ int main( int argc, char** argv )
             flip( view, view, 0 );
 
         vector<Point2f> pointbuf;
-        cvtColor(view, viewGray, CV_BGR2GRAY);
+        cvtColor(view, viewGray, COLOR_BGR2GRAY);
 
         bool found;
         switch( pattern )

diff --git a/samples/cpp/camshiftdemo.cpp b/samples/cpp/camshiftdemo.cpp
index 0b17239..8a1ed19 100644 (file)
--- a/samples/cpp/camshiftdemo.cpp
+++ b/samples/cpp/camshiftdemo.cpp
@@ -113,7 +113,7 @@ int main( int argc, const char** argv )
 
         if( !paused )
         {
-            cvtColor(image, hsv, CV_BGR2HSV);
+            cvtColor(image, hsv, COLOR_BGR2HSV);
 
             if( trackObject )
             {
@@ -163,7 +163,7 @@ int main( int argc, const char** argv )
                 }
 
                 if( backprojMode )
-                    cvtColor( backproj, image, CV_GRAY2BGR );
+                    cvtColor( backproj, image, COLOR_GRAY2BGR );
                 ellipse( image, trackBox, Scalar(0,0,255), 3, CV_AA );
             }
         }

diff --git a/samples/cpp/chamfer.cpp b/samples/cpp/chamfer.cpp
index 4be87cf..f83a24f 100644 (file)
--- a/samples/cpp/chamfer.cpp
+++ b/samples/cpp/chamfer.cpp
@@ -40,7 +40,7 @@ int main( int argc, const char** argv )
         return -1;
     }
     Mat cimg;
-    cvtColor(img, cimg, CV_GRAY2BGR);
+    cvtColor(img, cimg, COLOR_GRAY2BGR);
 
     // if the image and the template are not edge maps but normal grayscale images,
     // you might want to uncomment the lines below to produce the maps. You can also

diff --git a/samples/cpp/descriptor_extractor_matcher.cpp b/samples/cpp/descriptor_extractor_matcher.cpp
index 7aa5299..98c6452 100644 (file)
--- a/samples/cpp/descriptor_extractor_matcher.cpp
+++ b/samples/cpp/descriptor_extractor_matcher.cpp
@@ -208,7 +208,7 @@ static void doIteration( const Mat& img1, Mat& img2, bool isWarpPerspective,
                 matchesMask[i1] = 1;
         }
         // draw inliers
-        drawMatches( img1, keypoints1, img2, keypoints2, filteredMatches, drawImg, CV_RGB(0, 255, 0), CV_RGB(0, 0, 255), matchesMask
+        drawMatches( img1, keypoints1, img2, keypoints2, filteredMatches, drawImg, Scalar(0, 255, 0), Scalar(255, 0, 0), matchesMask
 #if DRAW_RICH_KEYPOINTS_MODE
                      , DrawMatchesFlags::DRAW_RICH_KEYPOINTS
 #endif
@@ -218,7 +218,7 @@ static void doIteration( const Mat& img1, Mat& img2, bool isWarpPerspective,
         // draw outliers
         for( size_t i1 = 0; i1 < matchesMask.size(); i1++ )
             matchesMask[i1] = !matchesMask[i1];
-        drawMatches( img1, keypoints1, img2, keypoints2, filteredMatches, drawImg, CV_RGB(0, 0, 255), CV_RGB(255, 0, 0), matchesMask,
+        drawMatches( img1, keypoints1, img2, keypoints2, filteredMatches, drawImg, Scalar(255, 0, 0), Scalar(0, 0, 255), matchesMask,
                      DrawMatchesFlags::DRAW_OVER_OUTIMG | DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS );
 #endif
 

diff --git a/samples/cpp/detection_based_tracker_sample.cpp b/samples/cpp/detection_based_tracker_sample.cpp
index 6aeb1b2..4c922de 100644 (file)
--- a/samples/cpp/detection_based_tracker_sample.cpp
+++ b/samples/cpp/detection_based_tracker_sample.cpp
@@ -116,7 +116,7 @@ static int test_FaceDetector(int argc, char *argv[])
         LOGD("\n\nSTEP n=%d        from prev step %f ms\n\n", n, t_ms);
         m=images[n-1];
         CV_Assert(! m.empty());
-        cvtColor(m, gray, CV_BGR2GRAY);
+        cvtColor(m, gray, COLOR_BGR2GRAY);
 
         fd.process(gray);
 

diff --git a/samples/cpp/distrans.cpp b/samples/cpp/distrans.cpp
index 4e3c3a2..e10a963 100644 (file)
--- a/samples/cpp/distrans.cpp
+++ b/samples/cpp/distrans.cpp
@@ -128,7 +128,7 @@ int main( int argc, const char** argv )
         // Call to update the view
         onTrackbar(0, 0);
 
-        int c = cvWaitKey(0) & 255;
+        int c = waitKey(0) & 255;
 
         if( c == 27 )
             break;

diff --git a/samples/cpp/drawing.cpp b/samples/cpp/drawing.cpp
index 5467463..313f100 100644 (file)
--- a/samples/cpp/drawing.cpp
+++ b/samples/cpp/drawing.cpp
@@ -157,14 +157,14 @@ int main()
             return 0;
     }
 
-    Size textsize = getTextSize("OpenCV forever!", CV_FONT_HERSHEY_COMPLEX, 3, 5, 0);
+    Size textsize = getTextSize("OpenCV forever!", FONT_HERSHEY_COMPLEX, 3, 5, 0);
     Point org((width - textsize.width)/2, (height - textsize.height)/2);
 
     Mat image2;
     for( i = 0; i < 255; i += 2 )
     {
         image2 = image - Scalar::all(i);
-        putText(image2, "OpenCV forever!", org, CV_FONT_HERSHEY_COMPLEX, 3,
+        putText(image2, "OpenCV forever!", org, FONT_HERSHEY_COMPLEX, 3,
                 Scalar(i, i, 255), 5, lineType);
 
         imshow(wndname, image2);

diff --git a/samples/cpp/edge.cpp b/samples/cpp/edge.cpp
index 6ff30a2..f11d20f 100644 (file)
--- a/samples/cpp/edge.cpp
+++ b/samples/cpp/edge.cpp
@@ -49,7 +49,7 @@ int main( int argc, const char** argv )
         return -1;
     }
     cedge.create(image.size(), image.type());
-    cvtColor(image, gray, CV_BGR2GRAY);
+    cvtColor(image, gray, COLOR_BGR2GRAY);
 
     // Create a window
     namedWindow("Edge map", 1);

diff --git a/samples/cpp/em.cpp b/samples/cpp/em.cpp
index 4c4664b..ab7f27e 100644 (file)
--- a/samples/cpp/em.cpp
+++ b/samples/cpp/em.cpp
@@ -59,7 +59,7 @@ int main( int /*argc*/, char** /*argv*/ )
     params.cov_mat_type = CvEM::COV_MAT_DIAGONAL;
     params.start_step = CvEM::START_E_STEP;
     params.means = em_model.get_means();
-    params.covs = (const CvMat**)em_model.get_covs();
+    params.covs = em_model.get_covs();
     params.weights = em_model.get_weights();
 
     em_model2.train( samples, Mat(), params, &labels );

diff --git a/samples/cpp/fback.cpp b/samples/cpp/fback.cpp
index 4923e8a..5293fd1 100644 (file)
--- a/samples/cpp/fback.cpp
+++ b/samples/cpp/fback.cpp
@@ -42,13 +42,13 @@ int main(int, char**)
     for(;;)
     {
         cap >> frame;
-        cvtColor(frame, gray, CV_BGR2GRAY);
+        cvtColor(frame, gray, COLOR_BGR2GRAY);
 
         if( prevgray.data )
         {
             calcOpticalFlowFarneback(prevgray, gray, flow, 0.5, 3, 15, 3, 5, 1.2, 0);
-            cvtColor(prevgray, cflow, CV_GRAY2BGR);
-            drawOptFlowMap(flow, cflow, 16, 1.5, CV_RGB(0, 255, 0));
+            cvtColor(prevgray, cflow, COLOR_GRAY2BGR);
+            drawOptFlowMap(flow, cflow, 16, 1.5, Scalar(0, 255, 0));
             imshow("flow", cflow);
         }
         if(waitKey(30)>=0)

diff --git a/samples/cpp/ffilldemo.cpp b/samples/cpp/ffilldemo.cpp
index a7fff77..fd697b2 100644 (file)
--- a/samples/cpp/ffilldemo.cpp
+++ b/samples/cpp/ffilldemo.cpp
@@ -81,7 +81,7 @@ int main( int argc, char** argv )
     }
     help();
     image0.copyTo(image);
-    cvtColor(image0, gray, CV_BGR2GRAY);
+    cvtColor(image0, gray, COLOR_BGR2GRAY);
     mask.create(image0.rows+2, image0.cols+2, CV_8UC1);
 
     namedWindow( "image", 0 );
@@ -106,7 +106,7 @@ int main( int argc, char** argv )
             if( isColor )
             {
                 cout << "Grayscale mode is set\n";
-                cvtColor(image0, gray, CV_BGR2GRAY);
+                cvtColor(image0, gray, COLOR_BGR2GRAY);
                 mask = Scalar::all(0);
                 isColor = false;
             }
@@ -135,7 +135,7 @@ int main( int argc, char** argv )
         case 'r':
             cout << "Original image is restored\n";
             image0.copyTo(image);
-            cvtColor(image, gray, CV_BGR2GRAY);
+            cvtColor(image, gray, COLOR_BGR2GRAY);
             mask = Scalar::all(0);
             break;
         case 's':

diff --git a/samples/cpp/generic_descriptor_match.cpp b/samples/cpp/generic_descriptor_match.cpp
index c86fdeb..1cb03c6 100644 (file)
--- a/samples/cpp/generic_descriptor_match.cpp
+++ b/samples/cpp/generic_descriptor_match.cpp
@@ -79,7 +79,7 @@ Mat DrawCorrespondences(const Mat& img1, const vector<KeyPoint>& features1, cons
 
     for (size_t i = 0; i < features1.size(); i++)
     {
-        circle(img_corr, features1[i].pt, 3, CV_RGB(255, 0, 0));
+        circle(img_corr, features1[i].pt, 3, Scalar(0, 0, 255));
     }
 
     for (size_t i = 0; i < features2.size(); i++)

diff --git a/samples/cpp/grabcut.cpp b/samples/cpp/grabcut.cpp
index 40280e4..b7d0d4d 100644 (file)
--- a/samples/cpp/grabcut.cpp
+++ b/samples/cpp/grabcut.cpp
@@ -296,15 +296,15 @@ int main( int argc, char** argv )
     help();
 
     const string winName = "image";
-    cvNamedWindow( winName.c_str(), CV_WINDOW_AUTOSIZE );
-    cvSetMouseCallback( winName.c_str(), on_mouse, 0 );
+    namedWindow( winName, WINDOW_AUTOSIZE );
+    setMouseCallback( winName, on_mouse, 0 );
 
     gcapp.setImageAndWinName( image, winName );
     gcapp.showImage();
 
     for(;;)
     {
-        int c = cvWaitKey(0);
+        int c = waitKey(0);
         switch( (char) c )
         {
         case '\x1b':
@@ -331,6 +331,6 @@ int main( int argc, char** argv )
     }
 
 exit_main:
-    cvDestroyWindow( winName.c_str() );
+    destroyWindow( winName );
     return 0;
 }

diff --git a/samples/cpp/houghcircles.cpp b/samples/cpp/houghcircles.cpp
index bc53fd1..a10e0d2 100644 (file)
--- a/samples/cpp/houghcircles.cpp
+++ b/samples/cpp/houghcircles.cpp
@@ -27,7 +27,7 @@ int main(int argc, char** argv)
 
     Mat cimg;
     medianBlur(img, img, 5);
-    cvtColor(img, cimg, CV_GRAY2BGR);
+    cvtColor(img, cimg, COLOR_GRAY2BGR);
 
     vector<Vec3f> circles;
     HoughCircles(img, circles, CV_HOUGH_GRADIENT, 1, 10,

diff --git a/samples/cpp/houghlines.cpp b/samples/cpp/houghlines.cpp
index 5ecc54c..681551a 100644 (file)
--- a/samples/cpp/houghlines.cpp
+++ b/samples/cpp/houghlines.cpp
@@ -27,7 +27,7 @@ int main(int argc, char** argv)
 
     Mat dst, cdst;
     Canny(src, dst, 50, 200, 3);
-    cvtColor(dst, cdst, CV_GRAY2BGR);
+    cvtColor(dst, cdst, COLOR_GRAY2BGR);
 
 #if 0
     vector<Vec2f> lines;

diff --git a/samples/cpp/image.cpp b/samples/cpp/image.cpp
index 5647f1c..41e2043 100644 (file)
--- a/samples/cpp/image.cpp
+++ b/samples/cpp/image.cpp
@@ -109,7 +109,7 @@ int main( int argc, char** argv )
     cvtColor(img_yuv, img, CV_YCrCb2BGR);
 
     // this is counterpart for cvNamedWindow
-    namedWindow("image with grain", CV_WINDOW_AUTOSIZE);
+    namedWindow("image with grain", WINDOW_AUTOSIZE);
 #if DEMO_MIXED_API_USE
     // this is to demonstrate that img and iplimg really share the data - the result of the above
     // processing is stored in img and thus in iplimg too.

diff --git a/samples/cpp/lkdemo.cpp b/samples/cpp/lkdemo.cpp
index bbb734c..90078f3 100644 (file)
--- a/samples/cpp/lkdemo.cpp
+++ b/samples/cpp/lkdemo.cpp
@@ -71,7 +71,7 @@ int main( int argc, char** argv )
             break;
 
         frame.copyTo(image);
-        cvtColor(image, gray, CV_BGR2GRAY);
+        cvtColor(image, gray, COLOR_BGR2GRAY);
 
         if( nightMode )
             image = Scalar::all(0);

diff --git a/samples/cpp/meanshift_segmentation.cpp b/samples/cpp/meanshift_segmentation.cpp
index ce34f6e..eb71619 100644 (file)
--- a/samples/cpp/meanshift_segmentation.cpp
+++ b/samples/cpp/meanshift_segmentation.cpp
@@ -65,7 +65,7 @@ int main(int argc, char** argv)
     colorRad = 10;
     maxPyrLevel = 1;
 
-    namedWindow( winName, CV_WINDOW_AUTOSIZE );
+    namedWindow( winName, WINDOW_AUTOSIZE );
 
     createTrackbar( "spatialRad", winName, &spatialRad, 80, meanShiftSegmentation );
     createTrackbar( "colorRad", winName, &colorRad, 60, meanShiftSegmentation );

diff --git a/samples/cpp/minarea.cpp b/samples/cpp/minarea.cpp
index 6056c39..c981c7f 100644 (file)
--- a/samples/cpp/minarea.cpp
+++ b/samples/cpp/minarea.cpp
@@ -54,7 +54,7 @@ int main( int /*argc*/, char** /*argv*/ )
 
         imshow( "rect & circle", img );
 
-        char key = (char)cvWaitKey();
+        char key = (char)waitKey();
         if( key == 27 || key == 'q' || key == 'Q' ) // 'ESC'
             break;
     }

diff --git a/samples/cpp/morphology2.cpp b/samples/cpp/morphology2.cpp
index d4ab351..330f4e0 100644 (file)
--- a/samples/cpp/morphology2.cpp
+++ b/samples/cpp/morphology2.cpp
@@ -79,7 +79,7 @@ int main( int argc, char** argv )
 
         OpenClose(open_close_pos, 0);
         ErodeDilate(erode_dilate_pos, 0);
-        c = cvWaitKey(0);
+        c = waitKey(0);
 
         if( (char)c == 27 )
             break;

diff --git a/samples/cpp/pca.cpp b/samples/cpp/pca.cpp
index ed23c76..be4e4a1 100644 (file)
--- a/samples/cpp/pca.cpp
+++ b/samples/cpp/pca.cpp
@@ -159,7 +159,7 @@ int main(int argc, char** argv)
 
     // init highgui window
     string winName = "Reconstruction | press 'q' to quit";
-    namedWindow(winName, CV_WINDOW_NORMAL);
+    namedWindow(winName, WINDOW_NORMAL);
 
     // params struct to pass to the trackbar handler
     params p;

diff --git a/samples/cpp/phase_corr.cpp b/samples/cpp/phase_corr.cpp
index e53f094..8267d2a 100644 (file)
--- a/samples/cpp/phase_corr.cpp
+++ b/samples/cpp/phase_corr.cpp
@@ -13,7 +13,7 @@ int main(int, char* [])
     do
     {
         video >> frame;
-        cvtColor(frame, curr, CV_RGB2GRAY);
+        cvtColor(frame, curr, COLOR_RGB2GRAY);
 
         if(prev.empty())
         {

diff --git a/samples/cpp/points_classifier.cpp b/samples/cpp/points_classifier.cpp
index 8bc35e5..e23b276 100644 (file)
--- a/samples/cpp/points_classifier.cpp
+++ b/samples/cpp/points_classifier.cpp
@@ -7,7 +7,7 @@
 using namespace std;
 using namespace cv;
 
-const Scalar WHITE_COLOR = CV_RGB(255,255,255);
+const Scalar WHITE_COLOR = Scalar(255,255,255);
 const string winName = "points";
 const int testStep = 5;
 
@@ -69,15 +69,15 @@ static void on_mouse( int event, int x, int y, int /*flags*/, void* )
         // put the text
         stringstream text;
         text << "current class " << classColors.size()-1;
-        putText( img, text.str(), Point(10,25), CV_FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
+        putText( img, text.str(), Point(10,25), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
 
         text.str("");
         text << "total classes " << classColors.size();
-        putText( img, text.str(), Point(10,50), CV_FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
+        putText( img, text.str(), Point(10,50), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
 
         text.str("");
         text << "total points " << trainedPoints.size();
-        putText(img, text.str(), cvPoint(10,75), CV_FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
+        putText(img, text.str(), Point(10,75), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
 
         // draw points
         for( size_t i = 0; i < trainedPoints.size(); i++ )
@@ -178,7 +178,7 @@ static void find_decision_boundary_SVM( CvSVMParams params )
     for( int i = 0; i < svmClassifier.get_support_vector_count(); i++ )
     {
         const float* supportVector = svmClassifier.get_support_vector(i);
-        circle( imgDst, Point(supportVector[0],supportVector[1]), 5, CV_RGB(255,255,255), -1 );
+        circle( imgDst, Point(supportVector[0],supportVector[1]), 5, Scalar(255,255,255), -1 );
     }
 
 }
@@ -526,7 +526,7 @@ int main()
         {
 #if _NBC_
             find_decision_boundary_NBC();
-            cvNamedWindow( "NormalBayesClassifier", WINDOW_AUTOSIZE );
+            namedWindow( "NormalBayesClassifier", WINDOW_AUTOSIZE );
             imshow( "NormalBayesClassifier", imgDst );
 #endif
 #if _KNN_
@@ -560,7 +560,7 @@ int main()
 
             params.C = 10;
             find_decision_boundary_SVM( params );
-            cvNamedWindow( "classificationSVM2", WINDOW_AUTOSIZE );
+            namedWindow( "classificationSVM2", WINDOW_AUTOSIZE );
             imshow( "classificationSVM2", imgDst );
 #endif
 

diff --git a/samples/cpp/rgbdodometry.cpp b/samples/cpp/rgbdodometry.cpp
index 660eb31..0db2fab 100644 (file)
--- a/samples/cpp/rgbdodometry.cpp
+++ b/samples/cpp/rgbdodometry.cpp
@@ -125,8 +125,8 @@ int main(int argc, char** argv)
     }
 
     Mat grayImage0, grayImage1, depthFlt0, depthFlt1/*in meters*/;
-    cvtColor( colorImage0, grayImage0, CV_BGR2GRAY );
-    cvtColor( colorImage1, grayImage1, CV_BGR2GRAY );
+    cvtColor( colorImage0, grayImage0, COLOR_BGR2GRAY );
+    cvtColor( colorImage1, grayImage1, COLOR_BGR2GRAY );
     depth0.convertTo( depthFlt0, CV_32FC1, 1./1000 );
     depth1.convertTo( depthFlt1, CV_32FC1, 1./1000 );
 

diff --git a/samples/cpp/segment_objects.cpp b/samples/cpp/segment_objects.cpp
index 8195df6..6438b89 100644 (file)
--- a/samples/cpp/segment_objects.cpp
+++ b/samples/cpp/segment_objects.cpp
@@ -95,8 +95,6 @@ int main(int argc, char** argv)
         if( !tmp_frame.data )
             break;
         bgsubtractor(tmp_frame, bgmask, update_bg_model ? -1 : 0);
-        //CvMat _bgmask = bgmask;
-        //cvSegmentFGMask(&_bgmask);
         refineSegments(tmp_frame, bgmask, out_frame);
         imshow("video", tmp_frame);
         imshow("segmented", out_frame);

diff --git a/samples/cpp/stereo_calib.cpp b/samples/cpp/stereo_calib.cpp
index 92135d6..809f23b 100644 (file)
--- a/samples/cpp/stereo_calib.cpp
+++ b/samples/cpp/stereo_calib.cpp
@@ -118,7 +118,7 @@ StereoCalib(const vector<string>& imagelist, Size boardSize, bool useCalibrated=
             {
                 cout << filename << endl;
                 Mat cimg, cimg1;
-                cvtColor(img, cimg, CV_GRAY2BGR);
+                cvtColor(img, cimg, COLOR_GRAY2BGR);
                 drawChessboardCorners(cimg, boardSize, corners, found);
                 double sf = 640./MAX(img.rows, img.cols);
                 resize(cimg, cimg1, Size(), sf, sf);
@@ -302,7 +302,7 @@ StereoCalib(const vector<string>& imagelist, Size boardSize, bool useCalibrated=
         {
             Mat img = imread(goodImageList[i*2+k], 0), rimg, cimg;
             remap(img, rimg, rmap[k][0], rmap[k][1], CV_INTER_LINEAR);
-            cvtColor(rimg, cimg, CV_GRAY2BGR);
+            cvtColor(rimg, cimg, COLOR_GRAY2BGR);
             Mat canvasPart = !isVerticalStereo ? canvas(Rect(w*k, 0, w, h)) : canvas(Rect(0, h*k, w, h));
             resize(cimg, canvasPart, canvasPart.size(), 0, 0, CV_INTER_AREA);
             if( useCalibrated )

diff --git a/samples/cpp/tutorial_code/HighGUI/BasicLinearTransformsTrackbar.cpp b/samples/cpp/tutorial_code/HighGUI/BasicLinearTransformsTrackbar.cpp
index b344c96..88c1dd4 100644 (file)
--- a/samples/cpp/tutorial_code/HighGUI/BasicLinearTransformsTrackbar.cpp
+++ b/samples/cpp/tutorial_code/HighGUI/BasicLinearTransformsTrackbar.cpp
@@ -24,17 +24,14 @@ Mat image;
  */
 static void on_trackbar( int, void* )
 {
-   Mat new_image = Mat::zeros( image.size(), image.type() );
+    Mat new_image = Mat::zeros( image.size(), image.type() );
 
-   for( int y = 0; y < image.rows; y++ )
-      { for( int x = 0; x < image.cols; x++ )
-           { for( int c = 0; c < 3; c++ )
-                {
-          new_image.at<Vec3b>(y,x)[c] = saturate_cast<uchar>( alpha*( image.at<Vec3b>(y,x)[c] ) + beta );
-                }
-       }
-      }
-   imshow("New Image", new_image);
+    for( int y = 0; y < image.rows; y++ )
+        for( int x = 0; x < image.cols; x++ )
+            for( int c = 0; c < 3; c++ )
+                new_image.at<Vec3b>(y,x)[c] = saturate_cast<uchar>( alpha*( image.at<Vec3b>(y,x)[c] ) + beta );
+
+    imshow("New Image", new_image);
 }
 
 

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/EqualizeHist_Demo.cpp b/samples/cpp/tutorial_code/Histograms_Matching/EqualizeHist_Demo.cpp
index 49e4be0..38a5839 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/EqualizeHist_Demo.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/EqualizeHist_Demo.cpp
@@ -31,14 +31,14 @@ int main( int, char** argv )
     }
 
   /// Convert to grayscale
-  cvtColor( src, src, CV_BGR2GRAY );
+  cvtColor( src, src, COLOR_BGR2GRAY );
 
   /// Apply Histogram Equalization
   equalizeHist( src, dst );
 
   /// Display results
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
-  namedWindow( equalized_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
+  namedWindow( equalized_window, WINDOW_AUTOSIZE );
 
   imshow( source_window, src );
   imshow( equalized_window, dst );

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/MatchTemplate_Demo.cpp b/samples/cpp/tutorial_code/Histograms_Matching/MatchTemplate_Demo.cpp
index e363357..ca318be 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/MatchTemplate_Demo.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/MatchTemplate_Demo.cpp
@@ -33,8 +33,8 @@ int main( int, char** argv )
   templ = imread( argv[2], 1 );
 
   /// Create windows
-  namedWindow( image_window, CV_WINDOW_AUTOSIZE );
-  namedWindow( result_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( image_window, WINDOW_AUTOSIZE );
+  namedWindow( result_window, WINDOW_AUTOSIZE );
 
   /// Create Trackbar
   const char* trackbar_label = "Method: \n 0: SQDIFF \n 1: SQDIFF NORMED \n 2: TM CCORR \n 3: TM CCORR NORMED \n 4: TM COEFF \n 5: TM COEFF NORMED";

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo1.cpp b/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo1.cpp
index f422603..86d2f2e 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo1.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo1.cpp
@@ -28,7 +28,7 @@ int main( int, char** argv )
   /// Read the image
   src = imread( argv[1], 1 );
   /// Transform it to HSV
-  cvtColor( src, hsv, CV_BGR2HSV );
+  cvtColor( src, hsv, COLOR_BGR2HSV );
 
   /// Use only the Hue value
   hue.create( hsv.size(), hsv.depth() );
@@ -37,7 +37,7 @@ int main( int, char** argv )
 
   /// Create Trackbar to enter the number of bins
   const char* window_image = "Source image";
-  namedWindow( window_image, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_image, WINDOW_AUTOSIZE );
   createTrackbar("* Hue  bins: ", window_image, &bins, 180, Hist_and_Backproj );
   Hist_and_Backproj(0, 0);
 

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo2.cpp b/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo2.cpp
index 42dd01a..85f18db 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo2.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/calcBackProject_Demo2.cpp
@@ -31,10 +31,10 @@ int main( int, char** argv )
   /// Read the image
   src = imread( argv[1], 1 );
   /// Transform it to HSV
-  cvtColor( src, hsv, CV_BGR2HSV );
+  cvtColor( src, hsv, COLOR_BGR2HSV );
 
   /// Show the image
-  namedWindow( window_image, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_image, WINDOW_AUTOSIZE );
   imshow( window_image, src );
 
   /// Set Trackbars for floodfill thresholds

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/calcHist_Demo.cpp b/samples/cpp/tutorial_code/Histograms_Matching/calcHist_Demo.cpp
index d3af5e7..577c8a8 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/calcHist_Demo.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/calcHist_Demo.cpp
@@ -71,7 +71,7 @@ int main( int, char** argv )
   }
 
   /// Display
-  namedWindow("calcHist Demo", CV_WINDOW_AUTOSIZE );
+  namedWindow("calcHist Demo", WINDOW_AUTOSIZE );
   imshow("calcHist Demo", histImage );
 
   waitKey(0);

diff --git a/samples/cpp/tutorial_code/Histograms_Matching/compareHist_Demo.cpp b/samples/cpp/tutorial_code/Histograms_Matching/compareHist_Demo.cpp
index b18aab2..f4dd4e5 100644 (file)
--- a/samples/cpp/tutorial_code/Histograms_Matching/compareHist_Demo.cpp
+++ b/samples/cpp/tutorial_code/Histograms_Matching/compareHist_Demo.cpp
@@ -22,24 +22,25 @@ int main( int argc, char** argv )
     Mat src_test2, hsv_test2;
     Mat hsv_half_down;
 
-   /// Load three images with different environment settings
-   if( argc < 4 )
-     { printf("** Error. Usage: ./compareHist_Demo <image_settings0> <image_setting1> <image_settings2>\n");
-       return -1;
-     }
+    /// Load three images with different environment settings
+    if( argc < 4 )
+    {
+        printf("** Error. Usage: ./compareHist_Demo <image_settings0> <image_setting1> <image_settings2>\n");
+        return -1;
+    }
 
-   src_base = imread( argv[1], 1 );
-   src_test1 = imread( argv[2], 1 );
-   src_test2 = imread( argv[3], 1 );
+    src_base = imread( argv[1], 1 );
+    src_test1 = imread( argv[2], 1 );
+    src_test2 = imread( argv[3], 1 );
 
-   /// Convert to HSV
-   cvtColor( src_base, hsv_base, CV_BGR2HSV );
-   cvtColor( src_test1, hsv_test1, CV_BGR2HSV );
-   cvtColor( src_test2, hsv_test2, CV_BGR2HSV );
+    /// Convert to HSV
+    cvtColor( src_base, hsv_base, COLOR_BGR2HSV );
+    cvtColor( src_test1, hsv_test1, COLOR_BGR2HSV );
+    cvtColor( src_test2, hsv_test2, COLOR_BGR2HSV );
 
-   hsv_half_down = hsv_base( Range( hsv_base.rows/2, hsv_base.rows - 1 ), Range( 0, hsv_base.cols - 1 ) );
+    hsv_half_down = hsv_base( Range( hsv_base.rows/2, hsv_base.rows - 1 ), Range( 0, hsv_base.cols - 1 ) );
 
-   /// Using 30 bins for hue and 32 for saturation
+    /// Using 30 bins for hue and 32 for saturation
     int h_bins = 50; int s_bins = 60;
     int histSize[] = { h_bins, s_bins };
 
@@ -74,14 +75,15 @@ int main( int argc, char** argv )
 
     /// Apply the histogram comparison methods
     for( int i = 0; i < 4; i++ )
-       { int compare_method = i;
-         double base_base = compareHist( hist_base, hist_base, compare_method );
-         double base_half = compareHist( hist_base, hist_half_down, compare_method );
-         double base_test1 = compareHist( hist_base, hist_test1, compare_method );
-         double base_test2 = compareHist( hist_base, hist_test2, compare_method );
-
-         printf( " Method [%d] Perfect, Base-Half, Base-Test(1), Base-Test(2) : %f, %f, %f, %f \n", i, base_base, base_half , base_test1, base_test2 );
-       }
+    {
+        int compare_method = i;
+        double base_base = compareHist( hist_base, hist_base, compare_method );
+        double base_half = compareHist( hist_base, hist_half_down, compare_method );
+        double base_test1 = compareHist( hist_base, hist_test1, compare_method );
+        double base_test2 = compareHist( hist_base, hist_test2, compare_method );
+
+        printf( " Method [%d] Perfect, Base-Half, Base-Test(1), Base-Test(2) : %f, %f, %f, %f \n", i, base_base, base_half , base_test1, base_test2 );
+    }
 
     printf( "Done \n" );
 

diff --git a/samples/cpp/tutorial_code/ImgProc/Morphology_1.cpp b/samples/cpp/tutorial_code/ImgProc/Morphology_1.cpp
index 47030fa..13a96a1 100644 (file)
--- a/samples/cpp/tutorial_code/ImgProc/Morphology_1.cpp
+++ b/samples/cpp/tutorial_code/ImgProc/Morphology_1.cpp
@@ -37,9 +37,9 @@ int main( int, char** argv )
     { return -1; }
 
   /// Create windows
-  namedWindow( "Erosion Demo", CV_WINDOW_AUTOSIZE );
-  namedWindow( "Dilation Demo", CV_WINDOW_AUTOSIZE );
-  cvMoveWindow( "Dilation Demo", src.cols, 0 );
+  namedWindow( "Erosion Demo", WINDOW_AUTOSIZE );
+  namedWindow( "Dilation Demo", WINDOW_AUTOSIZE );
+  moveWindow( "Dilation Demo", src.cols, 0 );
 
   /// Create Erosion Trackbar
   createTrackbar( "Element:\n 0: Rect \n 1: Cross \n 2: Ellipse", "Erosion Demo",

diff --git a/samples/cpp/tutorial_code/ImgProc/Morphology_2.cpp b/samples/cpp/tutorial_code/ImgProc/Morphology_2.cpp
index 2a1327d..a963bf8 100644 (file)
--- a/samples/cpp/tutorial_code/ImgProc/Morphology_2.cpp
+++ b/samples/cpp/tutorial_code/ImgProc/Morphology_2.cpp
@@ -39,20 +39,20 @@ int main( int, char** argv )
     { return -1; }
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Create Trackbar to select Morphology operation
   createTrackbar("Operator:\n 0: Opening - 1: Closing  \n 2: Gradient - 3: Top Hat \n 4: Black Hat", window_name, &morph_operator, max_operator, Morphology_Operations );
 
   /// Create Trackbar to select kernel type
   createTrackbar( "Element:\n 0: Rect - 1: Cross - 2: Ellipse", window_name,
-          &morph_elem, max_elem,
-          Morphology_Operations );
+                  &morph_elem, max_elem,
+                  Morphology_Operations );
 
   /// Create Trackbar to choose kernel size
   createTrackbar( "Kernel size:\n 2n +1", window_name,
-          &morph_size, max_kernel_size,
-          Morphology_Operations );
+                  &morph_size, max_kernel_size,
+                  Morphology_Operations );
 
   /// Default start
   Morphology_Operations( 0, 0 );

diff --git a/samples/cpp/tutorial_code/ImgProc/Pyramids.cpp b/samples/cpp/tutorial_code/ImgProc/Pyramids.cpp
index 0c90520..fc98d1c 100644 (file)
--- a/samples/cpp/tutorial_code/ImgProc/Pyramids.cpp
+++ b/samples/cpp/tutorial_code/ImgProc/Pyramids.cpp
@@ -40,7 +40,7 @@ int main( void )
   dst = tmp;
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
   imshow( window_name, dst );
 
   /// Loop

diff --git a/samples/cpp/tutorial_code/ImgProc/Smoothing.cpp b/samples/cpp/tutorial_code/ImgProc/Smoothing.cpp
index 5f51d07..8513bcf 100644 (file)
--- a/samples/cpp/tutorial_code/ImgProc/Smoothing.cpp
+++ b/samples/cpp/tutorial_code/ImgProc/Smoothing.cpp
@@ -31,7 +31,7 @@ int display_dst( int delay );
  */
 int main( void )
 {
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Load the source image
   src = imread( "../images/lena.png", 1 );
@@ -89,7 +89,7 @@ int display_caption( const char* caption )
   dst = Mat::zeros( src.size(), src.type() );
   putText( dst, caption,
            Point( src.cols/4, src.rows/2),
-           CV_FONT_HERSHEY_COMPLEX, 1, Scalar(255, 255, 255) );
+           FONT_HERSHEY_COMPLEX, 1, Scalar(255, 255, 255) );
 
   imshow( window_name, dst );
   int c = waitKey( DELAY_CAPTION );

diff --git a/samples/cpp/tutorial_code/ImgProc/Threshold.cpp b/samples/cpp/tutorial_code/ImgProc/Threshold.cpp
index 7505ec2..d98cc11 100644 (file)
--- a/samples/cpp/tutorial_code/ImgProc/Threshold.cpp
+++ b/samples/cpp/tutorial_code/ImgProc/Threshold.cpp
@@ -37,19 +37,19 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert the image to Gray
-  cvtColor( src, src_gray, CV_RGB2GRAY );
+  cvtColor( src, src_gray, COLOR_RGB2GRAY );
 
   /// Create a window to display results
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Create Trackbar to choose type of Threshold
   createTrackbar( trackbar_type,
-          window_name, &threshold_type,
-          max_type, Threshold_Demo );
+                  window_name, &threshold_type,
+                  max_type, Threshold_Demo );
 
   createTrackbar( trackbar_value,
-          window_name, &threshold_value,
-          max_value, Threshold_Demo );
+                  window_name, &threshold_value,
+                  max_value, Threshold_Demo );
 
   /// Call the function to initialize
   Threshold_Demo( 0, 0 );

diff --git a/samples/cpp/tutorial_code/ImgTrans/CannyDetector_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/CannyDetector_Demo.cpp
index c798f2f..7851c9f 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/CannyDetector_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/CannyDetector_Demo.cpp
@@ -58,10 +58,10 @@ int main( int, char** argv )
   dst.create( src.size(), src.type() );
 
   /// Convert the image to grayscale
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
   /// Create a window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Create a Trackbar for user to enter threshold
   createTrackbar( "Min Threshold:", window_name, &lowThreshold, max_lowThreshold, CannyThreshold );

diff --git a/samples/cpp/tutorial_code/ImgTrans/Geometric_Transforms_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/Geometric_Transforms_Demo.cpp
index 2d18357..0a2a2f5 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/Geometric_Transforms_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/Geometric_Transforms_Demo.cpp
@@ -65,13 +65,13 @@ int main( int, char** argv )
 
 
   /// Show what you got
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
-  namedWindow( warp_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( warp_window, WINDOW_AUTOSIZE );
   imshow( warp_window, warp_dst );
 
-  namedWindow( warp_rotate_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( warp_rotate_window, WINDOW_AUTOSIZE );
   imshow( warp_rotate_window, warp_rotate_dst );
 
   /// Wait until user exits the program

diff --git a/samples/cpp/tutorial_code/ImgTrans/HoughCircle_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/HoughCircle_Demo.cpp
index 71d37fb..da75026 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/HoughCircle_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/HoughCircle_Demo.cpp
@@ -25,7 +25,7 @@ int main(int, char** argv)
      { return -1; }
 
    /// Convert it to gray
-    cvtColor( src, src_gray, CV_BGR2GRAY );
+    cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
    /// Reduce the noise so we avoid false circle detection
     GaussianBlur( src_gray, src_gray, Size(9, 9), 2, 2 );
@@ -47,7 +47,7 @@ int main(int, char** argv)
     }
 
    /// Show your results
-    namedWindow( "Hough Circle Transform Demo", CV_WINDOW_AUTOSIZE );
+    namedWindow( "Hough Circle Transform Demo", WINDOW_AUTOSIZE );
     imshow( "Hough Circle Transform Demo", src );
 
     waitKey(0);

diff --git a/samples/cpp/tutorial_code/ImgTrans/HoughLines_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/HoughLines_Demo.cpp
index 561948a..ee30771 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/HoughLines_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/HoughLines_Demo.cpp
@@ -46,7 +46,7 @@ int main( int, char** argv )
      }
 
    /// Pass the image to gray
-   cvtColor( src, src_gray, CV_RGB2GRAY );
+   cvtColor( src, src_gray, COLOR_RGB2GRAY );
 
    /// Apply Canny edge detector
    Canny( src_gray, edges, 50, 200, 3 );
@@ -55,10 +55,10 @@ int main( int, char** argv )
    char thresh_label[50];
    sprintf( thresh_label, "Thres: %d + input", min_threshold );
 
-   namedWindow( standard_name, CV_WINDOW_AUTOSIZE );
+   namedWindow( standard_name, WINDOW_AUTOSIZE );
    createTrackbar( thresh_label, standard_name, &s_trackbar, max_trackbar, Standard_Hough);
 
-   namedWindow( probabilistic_name, CV_WINDOW_AUTOSIZE );
+   namedWindow( probabilistic_name, WINDOW_AUTOSIZE );
    createTrackbar( thresh_label, probabilistic_name, &p_trackbar, max_trackbar, Probabilistic_Hough);
 
    /// Initialize

diff --git a/samples/cpp/tutorial_code/ImgTrans/Laplace_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/Laplace_Demo.cpp
index 279dc6d..f6dff10 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/Laplace_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/Laplace_Demo.cpp
@@ -34,10 +34,10 @@ int main( int, char** argv )
   GaussianBlur( src, src, Size(3,3), 0, 0, BORDER_DEFAULT );
 
   /// Convert the image to grayscale
-  cvtColor( src, src_gray, CV_RGB2GRAY );
+  cvtColor( src, src_gray, COLOR_RGB2GRAY );
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Apply Laplace function
   Mat abs_dst;

diff --git a/samples/cpp/tutorial_code/ImgTrans/Remap_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/Remap_Demo.cpp
index 40e5118..66ed0f3 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/Remap_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/Remap_Demo.cpp
@@ -34,7 +34,7 @@ int main( int, char** argv )
   map_y.create( src.size(), CV_32FC1 );
 
   /// Create window
-  namedWindow( remap_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( remap_window, WINDOW_AUTOSIZE );
 
   /// Loop
   for(;;)

diff --git a/samples/cpp/tutorial_code/ImgTrans/Sobel_Demo.cpp b/samples/cpp/tutorial_code/ImgTrans/Sobel_Demo.cpp
index cbd03e8..f8c97c4 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/Sobel_Demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/Sobel_Demo.cpp
@@ -33,10 +33,10 @@ int main( int, char** argv )
   GaussianBlur( src, src, Size(3,3), 0, 0, BORDER_DEFAULT );
 
   /// Convert it to gray
-  cvtColor( src, src_gray, CV_RGB2GRAY );
+  cvtColor( src, src_gray, COLOR_RGB2GRAY );
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Generate grad_x and grad_y
   Mat grad_x, grad_y;

diff --git a/samples/cpp/tutorial_code/ImgTrans/copyMakeBorder_demo.cpp b/samples/cpp/tutorial_code/ImgTrans/copyMakeBorder_demo.cpp
index 0d84e73..0a441cc 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/copyMakeBorder_demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/copyMakeBorder_demo.cpp
@@ -43,7 +43,7 @@ int main( int, char** argv )
   printf( " ** Press 'ESC' to exit the program \n");
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Initialize arguments for the filter
   top = (int) (0.05*src.rows); bottom = (int) (0.05*src.rows);

diff --git a/samples/cpp/tutorial_code/ImgTrans/filter2D_demo.cpp b/samples/cpp/tutorial_code/ImgTrans/filter2D_demo.cpp
index 7a8dc76..86db4d8 100644 (file)
--- a/samples/cpp/tutorial_code/ImgTrans/filter2D_demo.cpp
+++ b/samples/cpp/tutorial_code/ImgTrans/filter2D_demo.cpp
@@ -35,7 +35,7 @@ int main ( int, char** argv )
     { return -1; }
 
   /// Create window
-  namedWindow( window_name, CV_WINDOW_AUTOSIZE );
+  namedWindow( window_name, WINDOW_AUTOSIZE );
 
   /// Initialize arguments for the filter
   anchor = Point( -1, -1 );

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/findContours_demo.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/findContours_demo.cpp
index e301476..bff125f 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/findContours_demo.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/findContours_demo.cpp
@@ -30,12 +30,12 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert image to gray and blur it
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
   blur( src_gray, src_gray, Size(3,3) );
 
   /// Create Window
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
   createTrackbar( " Canny thresh:", "Source", &thresh, max_thresh, thresh_callback );
@@ -68,6 +68,6 @@ void thresh_callback(int, void* )
      }
 
   /// Show in a window
-  namedWindow( "Contours", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Contours", WINDOW_AUTOSIZE );
   imshow( "Contours", drawing );
 }

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo1.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo1.cpp
index b973cfd..a752922 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo1.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo1.cpp
@@ -30,12 +30,12 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert image to gray and blur it
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
   blur( src_gray, src_gray, Size(3,3) );
 
   /// Create Window
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
   createTrackbar( " Threshold:", "Source", &thresh, max_thresh, thresh_callback );
@@ -83,6 +83,6 @@ void thresh_callback(int, void* )
      }
 
   /// Show in a window
-  namedWindow( "Contours", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Contours", WINDOW_AUTOSIZE );
   imshow( "Contours", drawing );
 }

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo2.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo2.cpp
index 70d8663..f21b718 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo2.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/generalContours_demo2.cpp
@@ -30,12 +30,12 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert image to gray and blur it
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
   blur( src_gray, src_gray, Size(3,3) );
 
   /// Create Window
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
   createTrackbar( " Threshold:", "Source", &thresh, max_thresh, thresh_callback );
@@ -85,6 +85,6 @@ void thresh_callback(int, void* )
      }
 
   /// Show in a window
-  namedWindow( "Contours", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Contours", WINDOW_AUTOSIZE );
   imshow( "Contours", drawing );
 }

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/hull_demo.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/hull_demo.cpp
index 8fe5d5b..e38003e 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/hull_demo.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/hull_demo.cpp
@@ -30,12 +30,12 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert image to gray and blur it
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
   blur( src_gray, src_gray, Size(3,3) );
 
   /// Create Window
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
   createTrackbar( " Threshold:", "Source", &thresh, max_thresh, thresh_callback );
@@ -62,7 +62,7 @@ void thresh_callback(int, void* )
   findContours( threshold_output, contours, hierarchy, CV_RETR_TREE, CV_CHAIN_APPROX_SIMPLE, Point(0, 0) );
 
   /// Find the convex hull object for each contour
- vector<vector<Point> >hull( contours.size() );
+  vector<vector<Point> >hull( contours.size() );
   for( size_t i = 0; i < contours.size(); i++ )
      {   convexHull( Mat(contours[i]), hull[i], false ); }
 
@@ -76,6 +76,6 @@ void thresh_callback(int, void* )
      }
 
   /// Show in a window
-  namedWindow( "Hull demo", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Hull demo", WINDOW_AUTOSIZE );
   imshow( "Hull demo", drawing );
 }

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/moments_demo.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/moments_demo.cpp
index 941edcc..9f0d286 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/moments_demo.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/moments_demo.cpp
@@ -30,12 +30,12 @@ int main( int, char** argv )
   src = imread( argv[1], 1 );
 
   /// Convert image to gray and blur it
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
   blur( src_gray, src_gray, Size(3,3) );
 
   /// Create Window
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
 
   createTrackbar( " Canny thresh:", "Source", &thresh, max_thresh, thresh_callback );
@@ -79,7 +79,7 @@ void thresh_callback(int, void* )
      }
 
   /// Show in a window
-  namedWindow( "Contours", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Contours", WINDOW_AUTOSIZE );
   imshow( "Contours", drawing );
 
   /// Calculate the area with the moments 00 and compare with the result of the OpenCV function

diff --git a/samples/cpp/tutorial_code/ShapeDescriptors/pointPolygonTest_demo.cpp b/samples/cpp/tutorial_code/ShapeDescriptors/pointPolygonTest_demo.cpp
index 25195d3..f55f8f6 100644 (file)
--- a/samples/cpp/tutorial_code/ShapeDescriptors/pointPolygonTest_demo.cpp
+++ b/samples/cpp/tutorial_code/ShapeDescriptors/pointPolygonTest_demo.cpp
@@ -71,9 +71,9 @@ int main( void )
 
   /// Create Window and show your results
   const char* source_window = "Source";
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, src );
-  namedWindow( "Distance", CV_WINDOW_AUTOSIZE );
+  namedWindow( "Distance", WINDOW_AUTOSIZE );
   imshow( "Distance", drawing );
 
   waitKey(0);

diff --git a/samples/cpp/tutorial_code/TrackingMotion/cornerDetector_Demo.cpp b/samples/cpp/tutorial_code/TrackingMotion/cornerDetector_Demo.cpp
index 5a37da4..31c7004 100644 (file)
--- a/samples/cpp/tutorial_code/TrackingMotion/cornerDetector_Demo.cpp
+++ b/samples/cpp/tutorial_code/TrackingMotion/cornerDetector_Demo.cpp
@@ -40,7 +40,7 @@ int main( int, char** argv )
 {
   /// Load source image and convert it to gray
   src = imread( argv[1], 1 );
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
   /// Set some parameters
   int blockSize = 3; int apertureSize = 3;
@@ -64,7 +64,7 @@ int main( int, char** argv )
   minMaxLoc( Mc, &myHarris_minVal, &myHarris_maxVal, 0, 0, Mat() );
 
   /* Create Window and Trackbar */
-  namedWindow( myHarris_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( myHarris_window, WINDOW_AUTOSIZE );
   createTrackbar( " Quality Level:", myHarris_window, &myHarris_qualityLevel, max_qualityLevel, myHarris_function );
   myHarris_function( 0, 0 );
 
@@ -75,7 +75,7 @@ int main( int, char** argv )
   minMaxLoc( myShiTomasi_dst, &myShiTomasi_minVal, &myShiTomasi_maxVal, 0, 0, Mat() );
 
   /* Create Window and Trackbar */
-  namedWindow( myShiTomasi_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( myShiTomasi_window, WINDOW_AUTOSIZE );
   createTrackbar( " Quality Level:", myShiTomasi_window, &myShiTomasi_qualityLevel, max_qualityLevel, myShiTomasi_function );
   myShiTomasi_function( 0, 0 );
 

diff --git a/samples/cpp/tutorial_code/TrackingMotion/cornerHarris_Demo.cpp b/samples/cpp/tutorial_code/TrackingMotion/cornerHarris_Demo.cpp
index 667ad7e..e048f05 100644 (file)
--- a/samples/cpp/tutorial_code/TrackingMotion/cornerHarris_Demo.cpp
+++ b/samples/cpp/tutorial_code/TrackingMotion/cornerHarris_Demo.cpp
@@ -31,10 +31,10 @@ int main( int, char** argv )
 {
   /// Load source image and convert it to gray
   src = imread( argv[1], 1 );
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
   /// Create a window and a trackbar
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   createTrackbar( "Threshold: ", source_window, &thresh, max_thresh, cornerHarris_demo );
   imshow( source_window, src );
 
@@ -77,6 +77,6 @@ void cornerHarris_demo( int, void* )
           }
      }
   /// Showing the result
-  namedWindow( corners_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( corners_window, WINDOW_AUTOSIZE );
   imshow( corners_window, dst_norm_scaled );
 }

diff --git a/samples/cpp/tutorial_code/TrackingMotion/cornerSubPix_Demo.cpp b/samples/cpp/tutorial_code/TrackingMotion/cornerSubPix_Demo.cpp
index 45da0f9..f695a28 100644 (file)
--- a/samples/cpp/tutorial_code/TrackingMotion/cornerSubPix_Demo.cpp
+++ b/samples/cpp/tutorial_code/TrackingMotion/cornerSubPix_Demo.cpp
@@ -32,10 +32,10 @@ int main( int, char** argv )
 {
   /// Load source image and convert it to gray
   src = imread( argv[1], 1 );
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
   /// Create Window
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
 
   /// Create Trackbar to set the number of corners
   createTrackbar( "Max  corners:", source_window, &maxCorners, maxTrackbar, goodFeaturesToTrack_Demo );
@@ -87,7 +87,7 @@ void goodFeaturesToTrack_Demo( int, void* )
      { circle( copy, corners[i], r, Scalar(rng.uniform(0,255), rng.uniform(0,255), rng.uniform(0,255)), -1, 8, 0 ); }
 
   /// Show what you got
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, copy );
 
   /// Set the neeed parameters to find the refined corners

diff --git a/samples/cpp/tutorial_code/TrackingMotion/goodFeaturesToTrack_Demo.cpp b/samples/cpp/tutorial_code/TrackingMotion/goodFeaturesToTrack_Demo.cpp
index f20669f..b45d60a 100644 (file)
--- a/samples/cpp/tutorial_code/TrackingMotion/goodFeaturesToTrack_Demo.cpp
+++ b/samples/cpp/tutorial_code/TrackingMotion/goodFeaturesToTrack_Demo.cpp
@@ -32,10 +32,10 @@ int main( int, char** argv )
 {
   /// Load source image and convert it to gray
   src = imread( argv[1], 1 );
-  cvtColor( src, src_gray, CV_BGR2GRAY );
+  cvtColor( src, src_gray, COLOR_BGR2GRAY );
 
   /// Create Window
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
 
   /// Create Trackbar to set the number of corners
   createTrackbar( "Max  corners:", source_window, &maxCorners, maxTrackbar, goodFeaturesToTrack_Demo );
@@ -87,6 +87,6 @@ void goodFeaturesToTrack_Demo( int, void* )
      { circle( copy, corners[i], r, Scalar(rng.uniform(0,255), rng.uniform(0,255), rng.uniform(0,255)), -1, 8, 0 ); }
 
   /// Show what you got
-  namedWindow( source_window, CV_WINDOW_AUTOSIZE );
+  namedWindow( source_window, WINDOW_AUTOSIZE );
   imshow( source_window, copy );
 }

diff --git a/samples/cpp/tutorial_code/calib3d/camera_calibration/camera_calibration.cpp b/samples/cpp/tutorial_code/calib3d/camera_calibration/camera_calibration.cpp
index 5cc15b3..6df0c16 100644 (file)
--- a/samples/cpp/tutorial_code/calib3d/camera_calibration/camera_calibration.cpp
+++ b/samples/cpp/tutorial_code/calib3d/camera_calibration/camera_calibration.cpp
@@ -294,7 +294,7 @@ int main(int argc, char* argv[])
                 if( s.calibrationPattern == Settings::CHESSBOARD)
                 {
                     Mat viewGray;
-                    cvtColor(view, viewGray, CV_BGR2GRAY);
+                    cvtColor(view, viewGray, COLOR_BGR2GRAY);
                     cornerSubPix( viewGray, pointBuf, Size(11,11),
                         Size(-1,-1), TermCriteria( CV_TERMCRIT_EPS+CV_TERMCRIT_ITER, 30, 0.1 ));
                 }

diff --git a/samples/cpp/tutorial_code/calib3d/stereoBM/SBM_Sample.cpp b/samples/cpp/tutorial_code/calib3d/stereoBM/SBM_Sample.cpp
index 13d5aa8..9f87254 100644 (file)
--- a/samples/cpp/tutorial_code/calib3d/stereoBM/SBM_Sample.cpp
+++ b/samples/cpp/tutorial_code/calib3d/stereoBM/SBM_Sample.cpp
@@ -56,7 +56,7 @@ int main( int argc, char** argv )
   //-- 4. Display it as a CV_8UC1 image
   imgDisparity16S.convertTo( imgDisparity8U, CV_8UC1, 255/(maxVal - minVal));
 
-  namedWindow( windowDisparity, CV_WINDOW_NORMAL );
+  namedWindow( windowDisparity, WINDOW_NORMAL );
   imshow( windowDisparity, imgDisparity8U );
 
   //-- 5. Save the image

diff --git a/samples/cpp/tutorial_code/core/Matrix/Drawing_1.cpp b/samples/cpp/tutorial_code/core/Matrix/Drawing_1.cpp
index ef466e5..5cad992 100644 (file)
--- a/samples/cpp/tutorial_code/core/Matrix/Drawing_1.cpp
+++ b/samples/cpp/tutorial_code/core/Matrix/Drawing_1.cpp
@@ -64,9 +64,9 @@ int main( void ){
 
   /// 3. Display your stuff!
   imshow( atom_window, atom_image );
-  cvMoveWindow( atom_window, 0, 200 );
+  moveWindow( atom_window, 0, 200 );
   imshow( rook_window, rook_image );
-  cvMoveWindow( rook_window, w, 200 );
+  moveWindow( rook_window, w, 200 );
 
   waitKey( 0 );
   return(0);

diff --git a/samples/cpp/tutorial_code/core/Matrix/Drawing_2.cpp b/samples/cpp/tutorial_code/core/Matrix/Drawing_2.cpp
index c84eccb..844bcd2 100644 (file)
--- a/samples/cpp/tutorial_code/core/Matrix/Drawing_2.cpp
+++ b/samples/cpp/tutorial_code/core/Matrix/Drawing_2.cpp
@@ -304,7 +304,7 @@ int Displaying_Random_Text( Mat image, char* window_name, RNG rng )
  */
 int Displaying_Big_End( Mat image, char* window_name, RNG )
 {
-  Size textsize = getTextSize("OpenCV forever!", CV_FONT_HERSHEY_COMPLEX, 3, 5, 0);
+  Size textsize = getTextSize("OpenCV forever!", FONT_HERSHEY_COMPLEX, 3, 5, 0);
   Point org((window_width - textsize.width)/2, (window_height - textsize.height)/2);
   int lineType = 8;
 
@@ -313,7 +313,7 @@ int Displaying_Big_End( Mat image, char* window_name, RNG )
   for( int i = 0; i < 255; i += 2 )
   {
     image2 = image - Scalar::all(i);
-    putText( image2, "OpenCV forever!", org, CV_FONT_HERSHEY_COMPLEX, 3,
+    putText( image2, "OpenCV forever!", org, FONT_HERSHEY_COMPLEX, 3,
              Scalar(i, i, 255), 5, lineType );
 
     imshow( window_name, image2 );

diff --git a/samples/cpp/tutorial_code/core/interoperability_with_OpenCV_1/interoperability_with_OpenCV_1.cpp b/samples/cpp/tutorial_code/core/interoperability_with_OpenCV_1/interoperability_with_OpenCV_1.cpp
index aeeaf21..4a9be8d 100644 (file)
--- a/samples/cpp/tutorial_code/core/interoperability_with_OpenCV_1/interoperability_with_OpenCV_1.cpp
+++ b/samples/cpp/tutorial_code/core/interoperability_with_OpenCV_1/interoperability_with_OpenCV_1.cpp
@@ -45,7 +45,7 @@ int main( int argc, char** argv )
 
     // convert image to YUV color space. The output image will be created automatically.
     Mat I_YUV;
-    cvtColor(I, I_YUV, CV_BGR2YCrCb);
+    cvtColor(I, I_YUV, COLOR_BGR2YCrCb);
 
     vector<Mat> planes;    // Use the STL's vector structure to store multiple Mat objects
     split(I_YUV, planes);  // split the image into separate color planes (Y U V)
@@ -117,7 +117,7 @@ int main( int argc, char** argv )
     cvtColor(I_YUV, I, CV_YCrCb2BGR);  // and produce the output RGB image
 
 
-    namedWindow("image with grain", CV_WINDOW_AUTOSIZE);   // use this to create images
+    namedWindow("image with grain", WINDOW_AUTOSIZE);   // use this to create images
 
 #ifdef DEMO_MIXED_API_USE
     // this is to demonstrate that I and IplI really share the data - the result of the above

diff --git a/samples/cpp/tutorial_code/core/mat_mask_operations/mat_mask_operations.cpp b/samples/cpp/tutorial_code/core/mat_mask_operations/mat_mask_operations.cpp
index 555d668..a446cfd 100644 (file)
--- a/samples/cpp/tutorial_code/core/mat_mask_operations/mat_mask_operations.cpp
+++ b/samples/cpp/tutorial_code/core/mat_mask_operations/mat_mask_operations.cpp
@@ -30,8 +30,8 @@ int main( int argc, char* argv[])
     else
         I = imread( filename, CV_LOAD_IMAGE_COLOR);
 
-    namedWindow("Input", CV_WINDOW_AUTOSIZE);
-    namedWindow("Output", CV_WINDOW_AUTOSIZE);
+    namedWindow("Input", WINDOW_AUTOSIZE);
+    namedWindow("Output", WINDOW_AUTOSIZE);
 
     imshow("Input", I);
     double t = (double)getTickCount();
@@ -42,7 +42,7 @@ int main( int argc, char* argv[])
     cout << "Hand written function times passed in seconds: " << t << endl;
 
     imshow("Output", J);
-    cvWaitKey(0);
+    waitKey(0);
 
     Mat kern = (Mat_<char>(3,3) <<  0, -1,  0,
                                    -1,  5, -1,
@@ -54,7 +54,7 @@ int main( int argc, char* argv[])
 
     imshow("Output", K);
 
-    cvWaitKey(0);
+    waitKey(0);
     return 0;
 }
 void Sharpen(const Mat& myImage,Mat& Result)

diff --git a/samples/cpp/tutorial_code/features2D/SURF_Homography.cpp b/samples/cpp/tutorial_code/features2D/SURF_Homography.cpp
index 506e3b4..4b29638 100644 (file)
--- a/samples/cpp/tutorial_code/features2D/SURF_Homography.cpp
+++ b/samples/cpp/tutorial_code/features2D/SURF_Homography.cpp
@@ -95,8 +95,8 @@ int main( int argc, char** argv )
 
   //-- Get the corners from the image_1 ( the object to be "detected" )
   std::vector<Point2f> obj_corners(4);
-  obj_corners[0] = cvPoint(0,0); obj_corners[1] = cvPoint( img_object.cols, 0 );
-  obj_corners[2] = cvPoint( img_object.cols, img_object.rows ); obj_corners[3] = cvPoint( 0, img_object.rows );
+  obj_corners[0] = Point(0,0); obj_corners[1] = Point( img_object.cols, 0 );
+  obj_corners[2] = Point( img_object.cols, img_object.rows ); obj_corners[3] = Point( 0, img_object.rows );
   std::vector<Point2f> scene_corners(4);
 
   perspectiveTransform( obj_corners, scene_corners, H);

diff --git a/samples/cpp/tutorial_code/introduction/display_image/display_image.cpp b/samples/cpp/tutorial_code/introduction/display_image/display_image.cpp
index 8990bac..ef8abfd 100644 (file)
--- a/samples/cpp/tutorial_code/introduction/display_image/display_image.cpp
+++ b/samples/cpp/tutorial_code/introduction/display_image/display_image.cpp
@@ -22,7 +22,7 @@ int main( int argc, char** argv )
         return -1;
     }
 
-    namedWindow( "Display window", CV_WINDOW_AUTOSIZE );// Create a window for display.
+    namedWindow( "Display window", WINDOW_AUTOSIZE );// Create a window for display.
     imshow( "Display window", image );                   // Show our image inside it.
 
     waitKey(0);                                                                                         // Wait for a keystroke in the window

diff --git a/samples/cpp/tutorial_code/introduction/windows_visual_studio_Opencv/Test.cpp b/samples/cpp/tutorial_code/introduction/windows_visual_studio_Opencv/Test.cpp
index cf35383..240a2e1 100644 (file)
--- a/samples/cpp/tutorial_code/introduction/windows_visual_studio_Opencv/Test.cpp
+++ b/samples/cpp/tutorial_code/introduction/windows_visual_studio_Opencv/Test.cpp
@@ -74,10 +74,10 @@ int main(int argc, char *argv[])
     const char* WIN_RF = "Reference";
 
     // Windows
-            namedWindow(WIN_RF, CV_WINDOW_AUTOSIZE );
-            namedWindow(WIN_UT, CV_WINDOW_AUTOSIZE );
-            cvMoveWindow(WIN_RF, 400       ,            0);             //750,  2 (bernat =0)
-            cvMoveWindow(WIN_UT, refS.width,            0);             //1500, 2
+            namedWindow(WIN_RF, WINDOW_AUTOSIZE );
+            namedWindow(WIN_UT, WINDOW_AUTOSIZE );
+            moveWindow(WIN_RF, 400       ,            0);               //750,  2 (bernat =0)
+            moveWindow(WIN_UT, refS.width,            0);               //1500, 2
 
     cout << "Frame resolution: Width=" << refS.width << "  Height=" << refS.height
          << " of nr#: " << captRefrnc.get(CV_CAP_PROP_FRAME_COUNT) << endl;
@@ -124,7 +124,7 @@ int main(int argc, char *argv[])
         imshow( WIN_RF, frameReference);
         imshow( WIN_UT, frameUnderTest);
 
-        c = (char)cvWaitKey(delay);
+        c = (char)waitKey(delay);
         if (c == 27) break;
     }
 

diff --git a/samples/cpp/tutorial_code/objectDetection/objectDetection.cpp b/samples/cpp/tutorial_code/objectDetection/objectDetection.cpp
index e20a7d4..c2465c4 100644 (file)
--- a/samples/cpp/tutorial_code/objectDetection/objectDetection.cpp
+++ b/samples/cpp/tutorial_code/objectDetection/objectDetection.cpp
@@ -30,7 +30,7 @@ RNG rng(12345);
  */
 int main( void )
 {
-  CvCapture* capture;
+  VideoCapture capture;
   Mat frame;
 
   //-- 1. Load the cascades
@@ -38,12 +38,12 @@ int main( void )
   if( !eyes_cascade.load( eyes_cascade_name ) ){ printf("--(!)Error loading\n"); return -1; };
 
   //-- 2. Read the video stream
-  capture = cvCaptureFromCAM( -1 );
-  if( capture )
+  capture.open( -1 );
+  if( capture.isOpened() )
   {
     for(;;)
     {
-      frame = cvQueryFrame( capture );
+      capture >> frame;
 
       //-- 3. Apply the classifier to the frame
       if( !frame.empty() )
@@ -67,7 +67,7 @@ void detectAndDisplay( Mat frame )
    std::vector<Rect> faces;
    Mat frame_gray;
 
-   cvtColor( frame, frame_gray, CV_BGR2GRAY );
+   cvtColor( frame, frame_gray, COLOR_BGR2GRAY );
    equalizeHist( frame_gray, frame_gray );
    //-- Detect faces
    face_cascade.detectMultiScale( frame_gray, faces, 1.1, 2, 0|CV_HAAR_SCALE_IMAGE, Size(30, 30) );

diff --git a/samples/cpp/tutorial_code/objectDetection/objectDetection2.cpp b/samples/cpp/tutorial_code/objectDetection/objectDetection2.cpp
index 75167f6..bfa1f15 100644 (file)
--- a/samples/cpp/tutorial_code/objectDetection/objectDetection2.cpp
+++ b/samples/cpp/tutorial_code/objectDetection/objectDetection2.cpp
@@ -30,7 +30,7 @@ RNG rng(12345);
  */
 int main( void )
 {
-  CvCapture* capture;
+  VideoCapture capture;
   Mat frame;
 
   //-- 1. Load the cascade
@@ -38,12 +38,12 @@ int main( void )
   if( !eyes_cascade.load( eyes_cascade_name ) ){ printf("--(!)Error loading\n"); return -1; };
 
   //-- 2. Read the video stream
-  capture = cvCaptureFromCAM( -1 );
-  if( capture )
+  capture.open( -1 );
+  if( capture.isOpened() )
   {
     for(;;)
     {
-      frame = cvQueryFrame( capture );
+      capture >> frame;
 
       //-- 3. Apply the classifier to the frame
       if( !frame.empty() )
@@ -67,7 +67,7 @@ void detectAndDisplay( Mat frame )
    std::vector<Rect> faces;
    Mat frame_gray;
 
-   cvtColor( frame, frame_gray, CV_BGR2GRAY );
+   cvtColor( frame, frame_gray, COLOR_BGR2GRAY );
    equalizeHist( frame_gray, frame_gray );
 
    //-- Detect faces

diff --git a/samples/cpp/video_dmtx.cpp b/samples/cpp/video_dmtx.cpp
index 4d36a38..01eadd4 100644 (file)
--- a/samples/cpp/video_dmtx.cpp
+++ b/samples/cpp/video_dmtx.cpp
@@ -52,7 +52,7 @@ namespace
             if (frame.empty())
                 break;
             cv::Mat gray;
-            cv::cvtColor(frame,gray,CV_RGB2GRAY);
+            cv::cvtColor(frame,gray,COLOR_RGB2GRAY);
             vector<string> codes;
             Mat corners;
             findDataMatrix(gray, codes, corners);

diff --git a/samples/cpp/video_homography.cpp b/samples/cpp/video_homography.cpp
index 01af565..c838800 100644 (file)
--- a/samples/cpp/video_homography.cpp
+++ b/samples/cpp/video_homography.cpp
@@ -161,7 +161,7 @@ int main(int ac, char ** av)
         if (frame.empty())
             break;
 
-        cvtColor(frame, gray, CV_RGB2GRAY);
+        cvtColor(frame, gray, COLOR_RGB2GRAY);
 
         detector.detect(gray, query_kpts); //Find interest points
 

diff --git a/samples/cpp/watershed.cpp b/samples/cpp/watershed.cpp
index de7f525..e251d5c 100644 (file)
--- a/samples/cpp/watershed.cpp
+++ b/samples/cpp/watershed.cpp
@@ -58,8 +58,8 @@ int main( int argc, char** argv )
     namedWindow( "image", 1 );
 
     img0.copyTo(img);
-    cvtColor(img, markerMask, CV_BGR2GRAY);
-    cvtColor(markerMask, imgGray, CV_GRAY2BGR);
+    cvtColor(img, markerMask, COLOR_BGR2GRAY);
+    cvtColor(markerMask, imgGray, COLOR_GRAY2BGR);
     markerMask = Scalar::all(0);
     imshow( "image", img );
     setMouseCallback( "image", onMouse, 0 );