fixed kernel compilation in imgproc module

Merge pull request #1841 from ilya-lavrenov:ocl_kernel_compilation

fixed kernel compilation

Merge pull request #1824 from vpisarev:ocl_experiments5

Merge pull request #1828 from SpecLad:merge-2.4

Merge pull request #1816 from SpecLad:no-more-epoch

Merge pull request #1830 from SpecLad:config-includes

fixed bug with cl_long size on Win32 (thanks to Aaron Kunze)

fixed failure in Tonemap test

In the example in OpenCVConfig.cmake.in, explicitly add the include paths.

After #1801, the include directories are no longer added automatically,
so we should show the user how to do that.

In OpenCVConfig.cmake.in, lowercased the commands in the example.

This is the preferred style nowadays, and more consistent with the
rest of the code.

Merge remote-tracking branch 'origin/2.4' into merge-2.4

Conflicts:
modules/ocl/include/opencv2/ocl/ocl.hpp
modules/ocl/src/arithm.cpp
modules/ocl/src/build_warps.cpp
modules/ocl/src/color.cpp
modules/ocl/src/haar.cpp
modules/ocl/src/imgproc.cpp
modules/ocl/src/split_merge.cpp
modules/ocl/test/test_color.cpp
samples/cpp/3calibration.cpp
samples/cpp/OpenEXRimages_HDR_Retina_toneMapping.cpp
samples/cpp/OpenEXRimages_HDR_Retina_toneMapping_video.cpp
samples/cpp/Qt_sample/main.cpp
samples/cpp/camshiftdemo.cpp
samples/cpp/descriptor_extractor_matcher.cpp
samples/cpp/distrans.cpp
samples/cpp/generic_descriptor_match.cpp
samples/cpp/grabcut.cpp
samples/cpp/morphology2.cpp
samples/cpp/segment_objects.cpp
samples/cpp/stereo_calib.cpp
samples/cpp/tutorial_code/Histograms_Matching/compareHist_Demo.cpp
samples/cpp/tutorial_code/core/mat_mask_operations/mat_mask_operations.cpp
samples/cpp/tutorial_code/introduction/display_image/display_image.cpp
samples/cpp/tutorial_code/introduction/windows_visual_studio_Opencv/Test.cpp
samples/cpp/tutorial_code/objectDetection/objectDetection.cpp
samples/cpp/tutorial_code/objectDetection/objectDetection2.cpp
samples/cpp/video_dmtx.cpp

added some diagnostic into Tonemap test

another attempt to fix compile errors

fixed some more compile errors

fixed few compile errors and doc build errors

removed unnecessary opencl kernels

updated patch to bring in the first functions with "transparent API"

Merge pull request #1811 from Nerei:fixed_some_build_errors

Merge pull request #1722 from StevenPuttemans:feature_1631_second

Merge pull request #1807 from KonstantinMatskevich:intel_bug

Merge pull request #1809 from ilya-lavrenov:ocl_resize_nn

fixed Split for device version equals 78712

fixing bugs for Intel platform CPU device

Merge pull request #1808 from alalek:ocl_fix_cvtcolor_xyz

Applied all fixes related to translating C to C++ code
Also fixed some typos and code alignment
Also adapted tutorial CPP samples
Fixed some identation problems

Viz code beautification

added viz tutorial 3 test

fixed compilation for vs2008

fixed all Viz warnings
moved some headers to precomp.hpp

fixed warnings

Fixed Viz compilation: removed CV_EXPORTS from cv::viz::Color. The macro forced instantiation of all Matx<double, 4, 1> constructors and triggered StaticAssert there.

Fixed compilation errors: removed using namespace from hpp file. This led to pulling cv::ACCESS_MASK to global namespace where conflict with the same name in 'windows.h'

compilation & warnings

ocl: fix cvtColor XYZ: variables on stack

generalized ocl::resize for all data types (INTER_NEAREST mode)

Switched OpenCV to a 3-component version number.

Epoch is gone; now the version component names in C++ and CMake
are consistent (except REVISION corresponds to PATCH).

Merge pull request #1802 from SpecLad:cmake-templates

Merge pull request #1803 from ilya-lavrenov:ocl_cvtColor

Merge pull request #1804 from alekcac:youtube_link_fix

Merge pull request #1801 from vrabaud:master

Merge pull request #1799 from SpecLad:avformat-53-6

Update widget_pose.rst

changed Youtube link to our channel

Made the generated file templates' naming more consistent.

The general convention is <output file name> + ".in".

do not use include_directories when calling find_package(OpenCV)
This is something that should be left to the user, that's the whole
point of OpenCV_INCLUDE_DIRS.

Fix building with FFmpeg 0.8.

avformat_find_stream_info was introduced in Libav avformat 53.3,
but it was only merged to FFmpeg in avformat 53.6.

In Libav avformat 53.3-53.5 av_find_stream_info is not removed
(only deprecated), so this shouldn't break building with that.

Merge pull request #1794 from vrabaud:2.4

Merge pull request #1796 from ilya-lavrenov:ocl_cvtColor

Merge pull request #1792 from ilya-lavrenov:ocl_warp_eps

Merge pull request #1790 from ilya-lavrenov:ocl_ref

cleaner fix

added RGBA <-> mRGBA

added HLS -> RGB[A][FULL] conversion

added HSV -> RGB[A][FULL] conversion

added RGB[A] -> HLS[FULL] conversion

Merge pull request #1788 from alalek:perf_test_skip

return when OpenCV is found

added RGB[A] -> HSV[FULL] conversion

Merge pull request #1786 from SpecLad:cap-images-file-pattern

Merge pull request #1779 from perping:integral_2.4

Merge pull request #1784 from SpecLad:ffmpeg-no-png

fixes http://code.opencv.org/issues/3367

ocl: perf tests: skip tests if we fail check for maxMemoryAllocSize

perf tests: allow to skip performance tests

refactored some functions from ocl arithm

added eps to ocl::buildWarpPerspectiveMaps perf test

Merge pull request #1785 from ilya-lavrenov:ocl_kern_warn

Merge pull request #1789 from ilya-lavrenov:ocl_minMaxLoc

Merge pull request #1791 from prattmic:limit

Add missing limits include

Without limits included, several CUDA related files fail to compile with
GCC on Ubuntu:

modules/cudaimgproc/src/hough_lines.cpp:136:9: error: ‘numeric_limits’ is not a member of ‘std’

fix haar

fix match_template and haar

fixed compilation of ocl::minMaxLoc for Intel device

In the image sequence capture, only search for the ordinal in the file name.

Searching in directory names can yield confusing results; e.g. if
the input is "jpeg2000/image1.jp2", it will infer the pattern
"jpeg%04d/image1.jp2", which is likely not what the user intended.

If the user really desires for the variable part to be in the
directory name, it can always use an explicit pattern.

kernel warnings on AMD

Merge pull request #1775 from ilya-lavrenov:ocl_cvtColor

fix linux warning.

fix haar

Merge pull request #1782 from ilya-lavrenov:ocl_adapBilat

Merge pull request #1766 from ilya-lavrenov:ocl_remap_linear

Merge pull request #1773 from ilya-lavrenov:ocl_buildWarpPerspectiveMaps

Merge pull request #1783 from SpecLad:cap-no-pattern

fix some bug about haar and match_template.

added RGB5x5 <-> Gray

Merge pull request #1781 from SpecLad:merge-2.4

Replaced the image used in the Highgui_Video.ffmpeg_image test.

Our prebuilt FFmpeg Windows binaries don't have PNG support enabled
(because that requires zlib), so that makes a PNG image a bad choice
for this test.

When FFmpeg doesn't support PNG, VideoCapture falls back to the
"image sequence" implementation, which doesn't work for single images.

removed 2000x2000, 4000x4000 from test sizes of ocl::adaptiveBilateralFilter

Fixed the "image sequence" capture not failing when a pattern isn't found.

at can't be a null pointer, so the condition was always false, and
a nonsensical pattern like "image.png%00d" was being inferred.

added a performance test for ocl::buildWarpPerspectiveMaps; moved warps to a separate file

Added a citation which used to be in the gpu module in 2.4.

Removed another usage of __func__, following #1763.

Merge pull request #1780 from ilya-lavrenov:ocl_m_pi

Merge remote-tracking branch 'origin/2.4' into merge-2.4

Conflicts:
.gitignore
modules/contrib/src/detection_based_tracker.cpp
modules/core/include/opencv2/core/core.hpp
modules/core/include/opencv2/core/internal.hpp
modules/core/src/gpumat.cpp
modules/core/src/opengl.cpp
modules/gpu/src/cuda/safe_call.hpp
modules/highgui/src/cap.cpp
modules/imgproc/include/opencv2/imgproc/imgproc.hpp
modules/ocl/doc/image_processing.rst
modules/ocl/include/opencv2/ocl/ocl.hpp
modules/ocl/perf/perf_haar.cpp
modules/ocl/src/haar.cpp
modules/ocl/src/imgproc.cpp
modules/ocl/src/kmeans.cpp
modules/ocl/src/svm.cpp
modules/ocl/test/test_objdetect.cpp
samples/ocl/adaptive_bilateral_filter.cpp

added RGB5x5 <-> RGB conversion

Merge pull request #1767 from prattmic:python3_asstring

fix warning.

Make Integral sum support cv_32f, sqsum support cv_64f.

replaced custom PI by OpenCL M_PI constant

added RGB[A] <-> BGR[A] conversion to ocl::cvtColor

added XYZ to RGB conversion to ocl::cvtColor