Commit Graph

62 Commits

Author SHA1 Message Date
Andrey Pavlenko
e6e817e613 Revert "Merge pull request #1779 from perping:integral_2.4"
This reverts commit 54ea5bbac7, reversing
changes made to 28e0d3d771.
2014-03-28 16:05:04 +04:00
krodyush
de431609db optimize Dx and Dy calcualtion to make it as single opencl kernel 2013-12-17 14:01:01 +04:00
Alexander Smorkalov
514e97223c Some hacks to build and run OpenCL on Qualcomm S800.
Some of functions were enabled on Qualcomm S800 by changing grid size;
OpenCL kernel grid size unification for different platfroms;
Test pass rate improvements by inclreasing threshold;
Some tests were disabled for Android;
run.py was adopted for devices with brackets in in name.
2013-11-22 16:13:38 +04:00
Ilya Lavrenov
198cd1a40d added area fast mode to ocl::resize 2013-11-19 13:43:50 +04:00
Ilya Lavrenov
f70d63e4c9 added INTER_AREA interpolation to ocl::resize 2013-11-18 18:28:49 +04:00
Ilya Lavrenov
cc237b7aba generalized ocl::resize for all data types (INTER_NEAREST mode) 2013-11-15 22:52:29 +04:00
Roman Donchenko
54ea5bbac7 Merge pull request #1779 from perping:integral_2.4 2013-11-13 15:53:42 +04:00
perping
73b34e3f85 fix warning. 2013-11-11 15:06:58 +08:00
perping
50579d2524 Make Integral sum support cv_32f, sqsum support cv_64f. 2013-11-11 14:31:02 +08:00
Ilya Lavrenov
760b718981 added CV_16SC2 && CV_16UC1 map types support to ocl::remap (INTER_LINEAR mode) 2013-11-08 01:42:48 +04:00
Ilya Lavrenov
fa15769f39 added CV_16SC2 && CV_16UC1 maps support to ocl::remap (nearest neighbour only) 2013-11-06 23:14:04 +04:00
Andrey Pavlenko
bf58049d11 Merge pull request #1724 from ilya-lavrenov:ocl_thresh 2013-10-31 13:55:04 +04:00
Ilya Lavrenov
1d5f5d2364 vectorized ocl::threshold for single channel images 2013-10-31 12:00:53 +04:00
Andrey Pavlenko
ef9f690598 Merge pull request #1709 from ilya-lavrenov:ocl_cornerharris 2013-10-31 00:22:37 +04:00
Roman Donchenko
d11921f1a4 Merge pull request #1675 from krodyush:opencl-optimization 2013-10-30 12:28:31 +04:00
Ilya Lavrenov
bf6b5ee925 fixed extrapolation for ocl::cornerMinEigenVal and ocl::cornerHarris; enabled CV_32FC1 for the first function 2013-10-29 19:59:21 +04:00
Ilya Lavrenov
1f7f9c9682 rewrote and generalized ocl::threshold 2013-10-28 18:11:24 +04:00
konstantin
973137c78e fix bugs on host and device sides for imgprog_sobel3.cl 2013-10-28 00:14:07 +04:00
konstantin
ddd2b106fb rename imageproc_sobel2.cl -> imageproc_sobel3.cl for consistency 2013-10-25 21:44:01 +04:00
Andrey Pavlenko
2b6fca68bf fixing typo 2013-10-25 18:00:46 +04:00
Ilya Lavrenov
5864895ec6 fixed ocl::copyMakeBorder 2013-10-23 23:52:05 +04:00
konstantin
2f62940a0e Added few opencl optimizations (as Intel platform codepath):
1. HaarDetetctor: repack nodes to reduce memory footprint
2. cornerMinEigVal: 4 ocl kernels are fused into 1 for sobel calculation
2013-10-23 20:38:11 +04:00
Ilya Lavrenov
4f68f35a78 added reflect, reflect101, replicate, wrap border types to ocl::remap 2013-10-19 23:38:50 +04:00
Ilya Lavrenov
b3e1a4598c enabled double in copyMakeBorder kernel 2013-10-18 16:46:51 +04:00
Ilya Lavrenov
2681cef5d7 rewrote ocl::remap 2013-10-18 16:41:09 +04:00
Ilya Lavrenov
3fe74ad124 fixed ocl::copyMakeBorder 2013-10-18 16:39:47 +04:00
Ilya Lavrenov
36342eb425 added ROI support to ocl::convolve 2013-10-18 16:36:11 +04:00
Ilya Lavrenov
60f9ba0c64 added ROI support to ocl::CLAHE 2013-10-18 16:31:52 +04:00
Ilya Lavrenov
da3b7c79b8 refactored filters in the ocl module; added an accuracy tests for ocl::medianFilter 2013-10-17 12:23:56 +04:00
Ilya Lavrenov
dfbea3518a refactored ocl/src/imgproc.cpp 2013-10-11 23:39:41 +04:00
Ilya Lavrenov
1f51e6c0de fixed kernel compilation warnings on MacOSX 2013-10-10 22:56:39 +04:00
Ilya Lavrenov
9d1636daa6 catching OpenCL double not supported exceptions 2013-10-10 12:45:55 +04:00
Alexander Alekhin
1d6327ba2a ocl: fix Context::CL_* usage 2013-10-08 12:23:46 +04:00
Ilya Lavrenov
10d60f99dc fixed ocl::integral and enabled perf test for it 2013-10-04 12:33:14 +04:00
Alexander Alekhin
dd9ff587dc ocl: file-based ProgramCache refactoring 2013-10-03 19:50:14 +04:00
Alexander Alekhin
b00f79ac5f ocl: move program names into opencl_kernels.hpp 2013-10-03 19:50:14 +04:00
Alexander Alekhin
e8d9ed8955 ocl: split initialization.cpp into 3 files: context, operations, programcache 2013-10-03 19:50:14 +04:00
Ilya Lavrenov
311a7233c2 removed comments from filtering.cpp and imgproc.cpp in ocl module 2013-09-24 13:46:08 +04:00
Ilya Lavrenov
58b84c2fc0 removed needless divUp 2013-09-16 15:48:30 +04:00
Roman Donchenko
f55740da70 Deleted all trailing whitespace. 2013-08-21 18:59:24 +04:00
peng xiao
fc64faa22e Remove empty cv::ocl::CLAHE class. 2013-06-27 10:06:37 +08:00
peng xiao
cc3fb3d182 Let CLAHE_OCL reuse cv::CLAHE abstract class. 2013-06-26 10:50:51 +08:00
Peng Xiao
1d8cd3a717 Add ocl CLACH implementation.
Test cases (accuracy and performance) are provided.
2013-06-10 18:37:48 +08:00
Vadim Pisarevsky
5a4efe8bcf Merge pull request #924 from pengx17:2.4_arithm_fix 2013-05-30 12:01:25 +04:00
peng xiao
1d0c283508 Fix a bug when pushing pointers of arguments into std::vector.
When argument pointers pushed into an vector and the pointers point to
address on stack, we need to make sure they are valid until kernels are
successfully flushed onto the queue.
2013-05-28 17:27:55 +08:00
peng xiao
b4a4a05bdc Add ocl's good features to track implementation.
Additional notes with this commit:
1. Add cornerHarris_dxdy and cornerMinEigenVal_dxdy to get
the interim dx and dy output of Sobel operator;
2. Add minMax_buf to allow user to reuse buffers in minMax;
3. Fix an error when either min or max pointer fed into minMax is NULL;
4. Corner sorter temporarily uses C++ STL's quick sort. A parallel
 selection sort in OpneCL is contained in the implementation but disabled
due to poor performance at the moment.
5. Accuracy test for ocl gfft.
2013-05-22 13:46:42 +08:00
yao
b386ea72aa use float when sum overflow 2013-04-23 17:23:05 +08:00
Andrey Kamaev
7b8ad4cb04 Refactor OpenCL initialization and allow to use ocl module witout explicit setup 2013-03-21 18:18:15 +04:00
yao
65bb4250a4 Fix a bug in addWeighted x64 mode
And a potential problem in remap
2013-03-01 18:08:53 +08:00
yao
f1a5c1328c remove the empty implementation when HAVE_OPENCL not defined 2013-02-23 20:50:03 +08:00