Vadim Pisarevsky
|
3c7f7de477
|
avoid possible problems with zero values in tone mapping algorithms (http://code.opencv.org/issues/4020)
|
10 years ago |
Vadim Pisarevsky
|
b37aaa8303
|
significantly improved parallel non-local means by using granularity parameter in parallel_for_ loop. Because the algorithm deals with sliding sums, it's essential that each thread has enough work to do, otherwise the algorithm gets higher theoretical complexity and thus there is no speedup comparing to 1-thread code (at best).
|
10 years ago |
Vadim Pisarevsky
|
ee11a2d266
|
fully implemented SSE and NEON cases of intrin.hpp; extended the HAL with some basic math functions
|
10 years ago |
Alexander Alekhin
|
d326c77007
|
fix build (related to PR #3814)
|
10 years ago |
Alexander Alekhin
|
32da602bf7
|
fix build (related to PR #3814)
|
10 years ago |
Erik Karlsson
|
01d3df0d00
|
Added normType parameter to fastNlMeansDenoisingMulti
|
10 years ago |
Erik Karlsson
|
70a64ebe72
|
Added test cases
|
10 years ago |
Erik Karlsson
|
5f8d688664
|
Added parameter normType to fastNlMeansDenoising
|
10 years ago |
Erik Karlsson
|
9fff7896c5
|
Removed fastNlMeansDenoisingColored[Multi]Abs
|
10 years ago |
Erik Karlsson
|
82c54104d6
|
Fix warnings on Win x64
|
10 years ago |
Erik Karlsson
|
812edb5fdc
|
Fixed bug, maxDist() should reurn int, not double
|
10 years ago |
Erik Karlsson
|
8aa0783995
|
Changed parameter type of fastNlMeansDenoising[Multi][Abs] from std::vector<float> to const std::vector<float>&
|
10 years ago |
Erik Karlsson
|
b471f9ee26
|
Fixed call to tegra::fastNlMeansDenoising
|
10 years ago |
Erik Karlsson
|
4b5753daea
|
Corrected documentation
|
10 years ago |
Erik Karlsson
|
a594a0677a
|
Cleanup
|
10 years ago |
Erik Karlsson
|
c44488629a
|
Changed parameters of fastNlMeansDenoising[Multi][Abs] from float * to std::vector<float>
|
10 years ago |
Erik Karlsson
|
21160137d4
|
Addition of test cases
|
10 years ago |
Erik Karlsson
|
cc8d94c6fc
|
Addition of per-channel h-values for fastNlMeansDenoising[Multi][Abs]
|
10 years ago |
Erik Karlsson
|
324fa26848
|
Refactoring of OpenCL implementation
|
10 years ago |
Erik Karlsson
|
41ffcc27dd
|
Added support for h = 0.0
|
10 years ago |
Erik Karlsson
|
1e82a67cc4
|
Additional refactoring
|
10 years ago |
Erik Karlsson
|
c41efe4e30
|
Refactoring in preparation for per-channel h-values
|
10 years ago |
Erik Karlsson
|
18be52c05b
|
Changed LUTs from IT to int
|
10 years ago |
Erik Karlsson
|
305cff36e2
|
Changed from IT to int for distance calculation
|
10 years ago |
Erik Karlsson
|
69eae13ff3
|
Changed pointers from sample_t * to uchar *. Rescaling psz accordingly.
|
10 years ago |
Erik Karlsson
|
d56d04e41b
|
Updated documentation
|
10 years ago |
Erik Karlsson
|
ac6771f975
|
Added test cases
|
10 years ago |
Erik Karlsson
|
ae08884854
|
Added support for 4-component input for fastNlMeansDenoising[Multi][Abs]
|
10 years ago |
Erik Karlsson
|
87760d13fb
|
Cleanup and addition of 4-component support for ocl_fastNlMeansDenoising
|
10 years ago |
Maksim Shabunin
|
37c74e38f4
|
Python support
|
10 years ago |
Maksim Shabunin
|
457123027e
|
Modified java wrapping mechanism
|
10 years ago |
Erik Karlsson
|
50bb14a0a8
|
Avoiding unnecessary copy by creating borders in place after RGB -> RGBA conversion
|
10 years ago |
Erik Karlsson
|
73663dcdd1
|
Added support for 16-bit input
|
10 years ago |
Erik Karlsson
|
3bde9e9365
|
Added test cases
|
10 years ago |
Erik Karlsson
|
a9ff335a89
|
Added OpenCL support for FastNlMeansDenoisingAbs
|
10 years ago |
Erik Karlsson
|
9f7cac8c59
|
Addtion of test cases for CV_8UC3
|
10 years ago |
Erik Karlsson
|
0fdb95e195
|
Refactoring and addition of CV_8UC3 to ocl_fastNlMeansDenoising
|
10 years ago |
Vladislav Vinogradov
|
cda6fed41f
|
move tegra namespace out of cv to prevent conflicts
|
10 years ago |
Vladislav Vinogradov
|
44e41baffe
|
use new functions before all tegra:: calls
|
10 years ago |
Erik Karlsson
|
8e7aff4486
|
Changed fastNlMeansDenoising and fastNlMeansDenoisingMulti back to sum of sq distances. Moved sq sum of abs distances to fastNlMeansDenoisingAbs and fastNlMeansDenoisingMultiAbs
|
10 years ago |
Erik Karlsson
|
e647b7c7e8
|
Calculating almost_dist2weight at full size to avoid bounds checking
|
10 years ago |
Erik Karlsson
|
baf266c29e
|
Changed from sum of squared differences to sum of abs differences
|
10 years ago |
Erik Karlsson
|
c339720af9
|
Preparation for 16-bit colored denoising. Currently not working due to cvtColor not supportint 16-bit Lab conversion.
|
10 years ago |
Erik Karlsson
|
584372bbf2
|
Fixed bounds checking
|
10 years ago |
Erik Karlsson
|
d588c717da
|
Using WEIGHT_THRESHOLD to limit table size. Still problematic with 16-bit and big h-values.
|
10 years ago |
Erik Karlsson
|
42db9e7153
|
Basic 16-bit implmentation of fastNlMeansDenoising. Table-based exponetiation leads to high memory footprint and loss of precision in 16-bit mode.
|
10 years ago |
Erik Karlsson
|
8368fb9ea8
|
Additional refactoring preparing for 16-bit implementation
|
10 years ago |
Erik Karlsson
|
e178294b49
|
Refactoring in preparation for 16-bit implementation of fastNlMeansDenoising
|
10 years ago |
Vladislav Vinogradov
|
14e15fb4f7
|
fix documentation warnings
|
10 years ago |
Vladislav Vinogradov
|
f48befc6f0
|
refactor CUDA part of photo module
|
10 years ago |