1 /*M///////////////////////////////////////////////////////////////////////////////////////
\r
3 // IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
\r
5 // By downloading, copying, installing or using the software you agree to this license.
\r
6 // If you do not agree to this license, do not download, install,
\r
7 // copy or use the software.
\r
10 // License Agreement
\r
11 // For Open Source Computer Vision Library
\r
13 // Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
\r
14 // Copyright (C) 2009, Willow Garage Inc., all rights reserved.
\r
15 // Third party copyrights are property of their respective owners.
\r
17 // Redistribution and use in source and binary forms, with or without modification,
\r
18 // are permitted provided that the following conditions are met:
\r
20 // * Redistribution's of source code must retain the above copyright notice,
\r
21 // this list of conditions and the following disclaimer.
\r
23 // * Redistribution's in binary form must reproduce the above copyright notice,
\r
24 // this list of conditions and the following disclaimer in the documentation
\r
25 // and/or other GpuMaterials provided with the distribution.
\r
27 // * The name of the copyright holders may not be used to endorse or promote products
\r
28 // derived from this software without specific prior written permission.
\r
30 // This software is provided by the copyright holders and contributors "as is" and
\r
31 // any express or implied warranties, including, but not limited to, the implied
\r
32 // warranties of merchantability and fitness for a particular purpose are disclaimed.
\r
33 // In no event shall the Intel Corporation or contributors be liable for any direct,
\r
34 // indirect, incidental, special, exemplary, or consequential damages
\r
35 // (including, but not limited to, procurement of substitute goods or services;
\r
36 // loss of use, data, or profits; or business interruption) however caused
\r
37 // and on any theory of liability, whether in contract, strict liability,
\r
38 // or tort (including negligence or otherwise) arising in any way out of
\r
39 // the use of this software, even if advised of the possibility of such damage.
\r
43 #ifndef __OPENCV_GPU_HPP__
\r
44 #define __OPENCV_GPU_HPP__
\r
47 #include "opencv2/core/core.hpp"
\r
48 #include "opencv2/imgproc/imgproc.hpp"
\r
49 #include "opencv2/objdetect/objdetect.hpp"
\r
50 #include "opencv2/features2d/features2d.hpp"
\r
51 #include "opencv2/gpu/gpumat.hpp"
\r
57 //////////////////////////////// Initialization & Info ////////////////////////
\r
59 //! This is the only function that do not throw exceptions if the library is compiled without Cuda.
\r
60 CV_EXPORTS int getCudaEnabledDeviceCount();
\r
62 //! Functions below throw cv::Expception if the library is compiled without Cuda.
\r
64 CV_EXPORTS void setDevice(int device);
\r
65 CV_EXPORTS int getDevice();
\r
67 //! Explicitly destroys and cleans up all resources associated with the current device in the current process.
\r
68 //! Any subsequent API call to this device will reinitialize the device.
\r
69 CV_EXPORTS void resetDevice();
\r
73 FEATURE_SET_COMPUTE_10 = 10,
\r
74 FEATURE_SET_COMPUTE_11 = 11,
\r
75 FEATURE_SET_COMPUTE_12 = 12,
\r
76 FEATURE_SET_COMPUTE_13 = 13,
\r
77 FEATURE_SET_COMPUTE_20 = 20,
\r
78 FEATURE_SET_COMPUTE_21 = 21,
\r
79 GLOBAL_ATOMICS = FEATURE_SET_COMPUTE_11,
\r
80 SHARED_ATOMICS = FEATURE_SET_COMPUTE_12,
\r
81 NATIVE_DOUBLE = FEATURE_SET_COMPUTE_13
\r
84 // Gives information about what GPU archs this OpenCV GPU module was
\r
86 class CV_EXPORTS TargetArchs
\r
89 static bool builtWith(FeatureSet feature_set);
\r
90 static bool has(int major, int minor);
\r
91 static bool hasPtx(int major, int minor);
\r
92 static bool hasBin(int major, int minor);
\r
93 static bool hasEqualOrLessPtx(int major, int minor);
\r
94 static bool hasEqualOrGreater(int major, int minor);
\r
95 static bool hasEqualOrGreaterPtx(int major, int minor);
\r
96 static bool hasEqualOrGreaterBin(int major, int minor);
\r
101 // Gives information about the given GPU
\r
102 class CV_EXPORTS DeviceInfo
\r
105 // Creates DeviceInfo object for the current GPU
\r
106 DeviceInfo() : device_id_(getDevice()) { query(); }
\r
108 // Creates DeviceInfo object for the given GPU
\r
109 DeviceInfo(int device_id) : device_id_(device_id) { query(); }
\r
111 string name() const { return name_; }
\r
113 // Return compute capability versions
\r
114 int majorVersion() const { return majorVersion_; }
\r
115 int minorVersion() const { return minorVersion_; }
\r
117 int multiProcessorCount() const { return multi_processor_count_; }
\r
119 size_t freeMemory() const;
\r
120 size_t totalMemory() const;
\r
122 // Checks whether device supports the given feature
\r
123 bool supports(FeatureSet feature_set) const;
\r
125 // Checks whether the GPU module can be run on the given device
\r
126 bool isCompatible() const;
\r
128 int deviceID() const { return device_id_; }
\r
132 void queryMemory(size_t& free_memory, size_t& total_memory) const;
\r
137 int multi_processor_count_;
\r
142 //////////////////////////////// Error handling ////////////////////////
\r
144 CV_EXPORTS void error(const char *error_string, const char *file, const int line, const char *func);
\r
145 CV_EXPORTS void nppError( int err, const char *file, const int line, const char *func);
\r
147 //////////////////////////////// CudaMem ////////////////////////////////
\r
148 // CudaMem is limited cv::Mat with page locked memory allocation.
\r
149 // Page locked memory is only needed for async and faster coping to GPU.
\r
150 // It is convertable to cv::Mat header without reference counting
\r
151 // so you can use it with other opencv functions.
\r
153 // Page-locks the matrix m memory and maps it for the device(s)
\r
154 CV_EXPORTS void registerPageLocked(Mat& m);
\r
155 // Unmaps the memory of matrix m, and makes it pageable again.
\r
156 CV_EXPORTS void unregisterPageLocked(Mat& m);
\r
158 class CV_EXPORTS CudaMem
\r
161 enum { ALLOC_PAGE_LOCKED = 1, ALLOC_ZEROCOPY = 2, ALLOC_WRITE_COMBINED = 4 };
\r
164 CudaMem(const CudaMem& m);
\r
166 CudaMem(int rows, int cols, int type, int _alloc_type = ALLOC_PAGE_LOCKED);
\r
167 CudaMem(Size size, int type, int alloc_type = ALLOC_PAGE_LOCKED);
\r
170 //! creates from cv::Mat with coping data
\r
171 explicit CudaMem(const Mat& m, int alloc_type = ALLOC_PAGE_LOCKED);
\r
175 CudaMem& operator = (const CudaMem& m);
\r
177 //! returns deep copy of the matrix, i.e. the data is copied
\r
178 CudaMem clone() const;
\r
180 //! allocates new matrix data unless the matrix already has specified size and type.
\r
181 void create(int rows, int cols, int type, int alloc_type = ALLOC_PAGE_LOCKED);
\r
182 void create(Size size, int type, int alloc_type = ALLOC_PAGE_LOCKED);
\r
184 //! decrements reference counter and released memory if needed.
\r
187 //! returns matrix header with disabled reference counting for CudaMem data.
\r
188 Mat createMatHeader() const;
\r
189 operator Mat() const;
\r
191 //! maps host memory into device address space and returns GpuMat header for it. Throws exception if not supported by hardware.
\r
192 GpuMat createGpuMatHeader() const;
\r
193 operator GpuMat() const;
\r
195 //returns if host memory can be mapperd to gpu address space;
\r
196 static bool canMapHostMemory();
\r
198 // Please see cv::Mat for descriptions
\r
199 bool isContinuous() const;
\r
200 size_t elemSize() const;
\r
201 size_t elemSize1() const;
\r
204 int channels() const;
\r
205 size_t step1() const;
\r
207 bool empty() const;
\r
210 // Please see cv::Mat for descriptions
\r
224 //////////////////////////////// CudaStream ////////////////////////////////
\r
225 // Encapculates Cuda Stream. Provides interface for async coping.
\r
226 // Passed to each function that supports async kernel execution.
\r
227 // Reference counting is enabled
\r
229 class CV_EXPORTS Stream
\r
235 Stream(const Stream&);
\r
236 Stream& operator=(const Stream&);
\r
238 bool queryIfComplete();
\r
239 void waitForCompletion();
\r
241 //! downloads asynchronously.
\r
242 // Warning! cv::Mat must point to page locked memory (i.e. to CudaMem data or to its subMat)
\r
243 void enqueueDownload(const GpuMat& src, CudaMem& dst);
\r
244 void enqueueDownload(const GpuMat& src, Mat& dst);
\r
246 //! uploads asynchronously.
\r
247 // Warning! cv::Mat must point to page locked memory (i.e. to CudaMem data or to its ROI)
\r
248 void enqueueUpload(const CudaMem& src, GpuMat& dst);
\r
249 void enqueueUpload(const Mat& src, GpuMat& dst);
\r
251 void enqueueCopy(const GpuMat& src, GpuMat& dst);
\r
253 void enqueueMemSet(GpuMat& src, Scalar val);
\r
254 void enqueueMemSet(GpuMat& src, Scalar val, const GpuMat& mask);
\r
256 // converts matrix type, ex from float to uchar depending on type
\r
257 void enqueueConvert(const GpuMat& src, GpuMat& dst, int type, double a = 1, double b = 0);
\r
259 static Stream& Null();
\r
261 operator bool() const;
\r
270 friend struct StreamAccessor;
\r
272 explicit Stream(Impl* impl);
\r
276 //////////////////////////////// Filter Engine ////////////////////////////////
\r
279 The Base Class for 1D or Row-wise Filters
\r
281 This is the base class for linear or non-linear filters that process 1D data.
\r
282 In particular, such filters are used for the "horizontal" filtering parts in separable filters.
\r
284 class CV_EXPORTS BaseRowFilter_GPU
\r
287 BaseRowFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
\r
288 virtual ~BaseRowFilter_GPU() {}
\r
289 virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
\r
294 The Base Class for Column-wise Filters
\r
296 This is the base class for linear or non-linear filters that process columns of 2D arrays.
\r
297 Such filters are used for the "vertical" filtering parts in separable filters.
\r
299 class CV_EXPORTS BaseColumnFilter_GPU
\r
302 BaseColumnFilter_GPU(int ksize_, int anchor_) : ksize(ksize_), anchor(anchor_) {}
\r
303 virtual ~BaseColumnFilter_GPU() {}
\r
304 virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
\r
309 The Base Class for Non-Separable 2D Filters.
\r
311 This is the base class for linear or non-linear 2D filters.
\r
313 class CV_EXPORTS BaseFilter_GPU
\r
316 BaseFilter_GPU(const Size& ksize_, const Point& anchor_) : ksize(ksize_), anchor(anchor_) {}
\r
317 virtual ~BaseFilter_GPU() {}
\r
318 virtual void operator()(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null()) = 0;
\r
324 The Base Class for Filter Engine.
\r
326 The class can be used to apply an arbitrary filtering operation to an image.
\r
327 It contains all the necessary intermediate buffers.
\r
329 class CV_EXPORTS FilterEngine_GPU
\r
332 virtual ~FilterEngine_GPU() {}
\r
334 virtual void apply(const GpuMat& src, GpuMat& dst, Rect roi = Rect(0,0,-1,-1), Stream& stream = Stream::Null()) = 0;
\r
337 //! returns the non-separable filter engine with the specified filter
\r
338 CV_EXPORTS Ptr<FilterEngine_GPU> createFilter2D_GPU(const Ptr<BaseFilter_GPU>& filter2D, int srcType, int dstType);
\r
340 //! returns the separable filter engine with the specified filters
\r
341 CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
\r
342 const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType);
\r
343 CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableFilter_GPU(const Ptr<BaseRowFilter_GPU>& rowFilter,
\r
344 const Ptr<BaseColumnFilter_GPU>& columnFilter, int srcType, int bufType, int dstType, GpuMat& buf);
\r
346 //! returns horizontal 1D box filter
\r
347 //! supports only CV_8UC1 source type and CV_32FC1 sum type
\r
348 CV_EXPORTS Ptr<BaseRowFilter_GPU> getRowSumFilter_GPU(int srcType, int sumType, int ksize, int anchor = -1);
\r
350 //! returns vertical 1D box filter
\r
351 //! supports only CV_8UC1 sum type and CV_32FC1 dst type
\r
352 CV_EXPORTS Ptr<BaseColumnFilter_GPU> getColumnSumFilter_GPU(int sumType, int dstType, int ksize, int anchor = -1);
\r
354 //! returns 2D box filter
\r
355 //! supports CV_8UC1 and CV_8UC4 source type, dst type must be the same as source type
\r
356 CV_EXPORTS Ptr<BaseFilter_GPU> getBoxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1, -1));
\r
358 //! returns box filter engine
\r
359 CV_EXPORTS Ptr<FilterEngine_GPU> createBoxFilter_GPU(int srcType, int dstType, const Size& ksize,
\r
360 const Point& anchor = Point(-1,-1));
\r
362 //! returns 2D morphological filter
\r
363 //! only MORPH_ERODE and MORPH_DILATE are supported
\r
364 //! supports CV_8UC1 and CV_8UC4 types
\r
365 //! kernel must have CV_8UC1 type, one rows and cols == ksize.width * ksize.height
\r
366 CV_EXPORTS Ptr<BaseFilter_GPU> getMorphologyFilter_GPU(int op, int type, const Mat& kernel, const Size& ksize,
\r
367 Point anchor=Point(-1,-1));
\r
369 //! returns morphological filter engine. Only MORPH_ERODE and MORPH_DILATE are supported.
\r
370 CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel,
\r
371 const Point& anchor = Point(-1,-1), int iterations = 1);
\r
372 CV_EXPORTS Ptr<FilterEngine_GPU> createMorphologyFilter_GPU(int op, int type, const Mat& kernel, GpuMat& buf,
\r
373 const Point& anchor = Point(-1,-1), int iterations = 1);
\r
375 //! returns 2D filter with the specified kernel
\r
376 //! supports CV_8UC1 and CV_8UC4 types
\r
377 CV_EXPORTS Ptr<BaseFilter_GPU> getLinearFilter_GPU(int srcType, int dstType, const Mat& kernel, const Size& ksize,
\r
378 Point anchor = Point(-1, -1));
\r
380 //! returns the non-separable linear filter engine
\r
381 CV_EXPORTS Ptr<FilterEngine_GPU> createLinearFilter_GPU(int srcType, int dstType, const Mat& kernel,
\r
382 const Point& anchor = Point(-1,-1));
\r
384 //! returns the primitive row filter with the specified kernel.
\r
385 //! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 source type.
\r
386 //! there are two version of algorithm: NPP and OpenCV.
\r
387 //! NPP calls when srcType == CV_8UC1 or srcType == CV_8UC4 and bufType == srcType,
\r
388 //! otherwise calls OpenCV version.
\r
389 //! NPP supports only BORDER_CONSTANT border type.
\r
390 //! OpenCV version supports only CV_32F as buffer depth and
\r
391 //! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
\r
392 CV_EXPORTS Ptr<BaseRowFilter_GPU> getLinearRowFilter_GPU(int srcType, int bufType, const Mat& rowKernel,
\r
393 int anchor = -1, int borderType = BORDER_DEFAULT);
\r
395 //! returns the primitive column filter with the specified kernel.
\r
396 //! supports only CV_8UC1, CV_8UC4, CV_16SC1, CV_16SC2, CV_32SC1, CV_32FC1 dst type.
\r
397 //! there are two version of algorithm: NPP and OpenCV.
\r
398 //! NPP calls when dstType == CV_8UC1 or dstType == CV_8UC4 and bufType == dstType,
\r
399 //! otherwise calls OpenCV version.
\r
400 //! NPP supports only BORDER_CONSTANT border type.
\r
401 //! OpenCV version supports only CV_32F as buffer depth and
\r
402 //! BORDER_REFLECT101, BORDER_REPLICATE and BORDER_CONSTANT border types.
\r
403 CV_EXPORTS Ptr<BaseColumnFilter_GPU> getLinearColumnFilter_GPU(int bufType, int dstType, const Mat& columnKernel,
\r
404 int anchor = -1, int borderType = BORDER_DEFAULT);
\r
406 //! returns the separable linear filter engine
\r
407 CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
\r
408 const Mat& columnKernel, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
\r
409 int columnBorderType = -1);
\r
410 CV_EXPORTS Ptr<FilterEngine_GPU> createSeparableLinearFilter_GPU(int srcType, int dstType, const Mat& rowKernel,
\r
411 const Mat& columnKernel, GpuMat& buf, const Point& anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT,
\r
412 int columnBorderType = -1);
\r
414 //! returns filter engine for the generalized Sobel operator
\r
415 CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize,
\r
416 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
417 CV_EXPORTS Ptr<FilterEngine_GPU> createDerivFilter_GPU(int srcType, int dstType, int dx, int dy, int ksize, GpuMat& buf,
\r
418 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
420 //! returns the Gaussian filter engine
\r
421 CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, double sigma1, double sigma2 = 0,
\r
422 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
423 CV_EXPORTS Ptr<FilterEngine_GPU> createGaussianFilter_GPU(int type, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
\r
424 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
426 //! returns maximum filter
\r
427 CV_EXPORTS Ptr<BaseFilter_GPU> getMaxFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
\r
429 //! returns minimum filter
\r
430 CV_EXPORTS Ptr<BaseFilter_GPU> getMinFilter_GPU(int srcType, int dstType, const Size& ksize, Point anchor = Point(-1,-1));
\r
432 //! smooths the image using the normalized box filter
\r
433 //! supports CV_8UC1, CV_8UC4 types
\r
434 CV_EXPORTS void boxFilter(const GpuMat& src, GpuMat& dst, int ddepth, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null());
\r
436 //! a synonym for normalized box filter
\r
437 static inline void blur(const GpuMat& src, GpuMat& dst, Size ksize, Point anchor = Point(-1,-1), Stream& stream = Stream::Null()) { boxFilter(src, dst, -1, ksize, anchor, stream); }
\r
439 //! erodes the image (applies the local minimum operator)
\r
440 CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
\r
441 CV_EXPORTS void erode(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf, Point anchor = Point(-1, -1), int iterations = 1, Stream& stream = Stream::Null());
\r
443 //! dilates the image (applies the local maximum operator)
\r
444 CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
\r
445 CV_EXPORTS void dilate(const GpuMat& src, GpuMat& dst, const Mat& kernel, GpuMat& buf, Point anchor = Point(-1, -1), int iterations = 1, Stream& stream = Stream::Null());
\r
447 //! applies an advanced morphological operation to the image
\r
448 CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, Point anchor = Point(-1, -1), int iterations = 1);
\r
449 CV_EXPORTS void morphologyEx(const GpuMat& src, GpuMat& dst, int op, const Mat& kernel, GpuMat& buf1, GpuMat& buf2, Point anchor = Point(-1, -1), int iterations = 1, Stream& stream = Stream::Null());
\r
451 //! applies non-separable 2D linear filter to the image
\r
452 CV_EXPORTS void filter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernel, Point anchor=Point(-1,-1), Stream& stream = Stream::Null());
\r
454 //! applies separable 2D linear filter to the image
\r
455 CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY,
\r
456 Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
457 CV_EXPORTS void sepFilter2D(const GpuMat& src, GpuMat& dst, int ddepth, const Mat& kernelX, const Mat& kernelY, GpuMat& buf,
\r
458 Point anchor = Point(-1,-1), int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
\r
460 //! applies generalized Sobel operator to the image
\r
461 CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, int ksize = 3, double scale = 1,
\r
462 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
463 CV_EXPORTS void Sobel(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, int ksize = 3, double scale = 1,
\r
464 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
\r
466 //! applies the vertical or horizontal Scharr operator to the image
\r
467 CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, double scale = 1,
\r
468 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
469 CV_EXPORTS void Scharr(const GpuMat& src, GpuMat& dst, int ddepth, int dx, int dy, GpuMat& buf, double scale = 1,
\r
470 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
\r
472 //! smooths the image using Gaussian filter.
\r
473 CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, double sigma1, double sigma2 = 0,
\r
474 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1);
\r
475 CV_EXPORTS void GaussianBlur(const GpuMat& src, GpuMat& dst, Size ksize, GpuMat& buf, double sigma1, double sigma2 = 0,
\r
476 int rowBorderType = BORDER_DEFAULT, int columnBorderType = -1, Stream& stream = Stream::Null());
\r
478 //! applies Laplacian operator to the image
\r
479 //! supports only ksize = 1 and ksize = 3
\r
480 CV_EXPORTS void Laplacian(const GpuMat& src, GpuMat& dst, int ddepth, int ksize = 1, double scale = 1, Stream& stream = Stream::Null());
\r
483 ////////////////////////////// Arithmetics ///////////////////////////////////
\r
485 //! transposes the matrix
\r
486 //! supports matrix with element size = 1, 4 and 8 bytes (CV_8UC1, CV_8UC4, CV_16UC2, CV_32FC1, etc)
\r
487 CV_EXPORTS void transpose(const GpuMat& src1, GpuMat& dst, Stream& stream = Stream::Null());
\r
489 //! reverses the order of the rows, columns or both in a matrix
\r
490 //! supports CV_8UC1, CV_8UC4 types
\r
491 CV_EXPORTS void flip(const GpuMat& a, GpuMat& b, int flipCode, Stream& stream = Stream::Null());
\r
493 //! transforms 8-bit unsigned integers using lookup table: dst(i)=lut(src(i))
\r
494 //! destination array will have the depth type as lut and the same channels number as source
\r
495 //! supports CV_8UC1, CV_8UC3 types
\r
496 CV_EXPORTS void LUT(const GpuMat& src, const Mat& lut, GpuMat& dst, Stream& stream = Stream::Null());
\r
498 //! makes multi-channel array out of several single-channel arrays
\r
499 CV_EXPORTS void merge(const GpuMat* src, size_t n, GpuMat& dst, Stream& stream = Stream::Null());
\r
501 //! makes multi-channel array out of several single-channel arrays
\r
502 CV_EXPORTS void merge(const vector<GpuMat>& src, GpuMat& dst, Stream& stream = Stream::Null());
\r
504 //! copies each plane of a multi-channel array to a dedicated array
\r
505 CV_EXPORTS void split(const GpuMat& src, GpuMat* dst, Stream& stream = Stream::Null());
\r
507 //! copies each plane of a multi-channel array to a dedicated array
\r
508 CV_EXPORTS void split(const GpuMat& src, vector<GpuMat>& dst, Stream& stream = Stream::Null());
\r
510 //! computes magnitude of complex (x(i).re, x(i).im) vector
\r
511 //! supports only CV_32FC2 type
\r
512 CV_EXPORTS void magnitude(const GpuMat& x, GpuMat& magnitude, Stream& stream = Stream::Null());
\r
514 //! computes squared magnitude of complex (x(i).re, x(i).im) vector
\r
515 //! supports only CV_32FC2 type
\r
516 CV_EXPORTS void magnitudeSqr(const GpuMat& x, GpuMat& magnitude, Stream& stream = Stream::Null());
\r
518 //! computes magnitude of each (x(i), y(i)) vector
\r
519 //! supports only floating-point source
\r
520 CV_EXPORTS void magnitude(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
\r
522 //! computes squared magnitude of each (x(i), y(i)) vector
\r
523 //! supports only floating-point source
\r
524 CV_EXPORTS void magnitudeSqr(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, Stream& stream = Stream::Null());
\r
526 //! computes angle (angle(i)) of each (x(i), y(i)) vector
\r
527 //! supports only floating-point source
\r
528 CV_EXPORTS void phase(const GpuMat& x, const GpuMat& y, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
\r
530 //! converts Cartesian coordinates to polar
\r
531 //! supports only floating-point source
\r
532 CV_EXPORTS void cartToPolar(const GpuMat& x, const GpuMat& y, GpuMat& magnitude, GpuMat& angle, bool angleInDegrees = false, Stream& stream = Stream::Null());
\r
534 //! converts polar coordinates to Cartesian
\r
535 //! supports only floating-point source
\r
536 CV_EXPORTS void polarToCart(const GpuMat& magnitude, const GpuMat& angle, GpuMat& x, GpuMat& y, bool angleInDegrees = false, Stream& stream = Stream::Null());
\r
539 //////////////////////////// Per-element operations ////////////////////////////////////
\r
541 //! adds one matrix to another (c = a + b)
\r
542 //! supports CV_8UC1, CV_8UC4, CV_32SC1, CV_32FC1 types
\r
543 CV_EXPORTS void add(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
\r
544 //! adds scalar to a matrix (c = a + s)
\r
545 //! supports CV_32FC1 and CV_32FC2 type
\r
546 CV_EXPORTS void add(const GpuMat& a, const Scalar& sc, GpuMat& c, Stream& stream = Stream::Null());
\r
548 //! subtracts one matrix from another (c = a - b)
\r
549 //! supports CV_8UC1, CV_8UC4, CV_32SC1, CV_32FC1 types
\r
550 CV_EXPORTS void subtract(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
\r
551 //! subtracts scalar from a matrix (c = a - s)
\r
552 //! supports CV_32FC1 and CV_32FC2 type
\r
553 CV_EXPORTS void subtract(const GpuMat& a, const Scalar& sc, GpuMat& c, Stream& stream = Stream::Null());
\r
555 //! computes element-wise product of the two arrays (c = a * b)
\r
556 //! supports CV_8UC1, CV_8UC4, CV_32SC1, CV_32FC1 types
\r
557 CV_EXPORTS void multiply(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
\r
558 //! multiplies matrix to a scalar (c = a * s)
\r
559 //! supports CV_32FC1 type
\r
560 CV_EXPORTS void multiply(const GpuMat& a, const Scalar& sc, GpuMat& c, Stream& stream = Stream::Null());
\r
562 //! computes element-wise quotient of the two arrays (c = a / b)
\r
563 //! supports CV_8UC1, CV_8UC4, CV_32SC1, CV_32FC1 types
\r
564 CV_EXPORTS void divide(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
\r
565 //! computes element-wise quotient of matrix and scalar (c = a / s)
\r
566 //! supports CV_32FC1 type
\r
567 CV_EXPORTS void divide(const GpuMat& a, const Scalar& sc, GpuMat& c, Stream& stream = Stream::Null());
\r
569 //! computes exponent of each matrix element (b = e**a)
\r
570 //! supports only CV_32FC1 type
\r
571 CV_EXPORTS void exp(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
\r
573 //! computes power of each matrix element:
\r
574 // (dst(i,j) = pow( src(i,j) , power), if src.type() is integer
\r
575 // (dst(i,j) = pow(fabs(src(i,j)), power), otherwise
\r
576 //! supports all, except depth == CV_64F
\r
577 CV_EXPORTS void pow(const GpuMat& src, double power, GpuMat& dst, Stream& stream = Stream::Null());
\r
579 //! computes natural logarithm of absolute value of each matrix element: b = log(abs(a))
\r
580 //! supports only CV_32FC1 type
\r
581 CV_EXPORTS void log(const GpuMat& a, GpuMat& b, Stream& stream = Stream::Null());
\r
583 //! computes element-wise absolute difference of two arrays (c = abs(a - b))
\r
584 //! supports CV_8UC1, CV_8UC4, CV_32SC1, CV_32FC1 types
\r
585 CV_EXPORTS void absdiff(const GpuMat& a, const GpuMat& b, GpuMat& c, Stream& stream = Stream::Null());
\r
586 //! computes element-wise absolute difference of array and scalar (c = abs(a - s))
\r
587 //! supports only CV_32FC1 type
\r
588 CV_EXPORTS void absdiff(const GpuMat& a, const Scalar& s, GpuMat& c, Stream& stream = Stream::Null());
\r
590 //! compares elements of two arrays (c = a <cmpop> b)
\r
591 //! supports CV_8UC4, CV_32FC1 types
\r
592 CV_EXPORTS void compare(const GpuMat& a, const GpuMat& b, GpuMat& c, int cmpop, Stream& stream = Stream::Null());
\r
594 //! performs per-elements bit-wise inversion
\r
595 CV_EXPORTS void bitwise_not(const GpuMat& src, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
\r
597 //! calculates per-element bit-wise disjunction of two arrays
\r
598 CV_EXPORTS void bitwise_or(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
\r
600 //! calculates per-element bit-wise conjunction of two arrays
\r
601 CV_EXPORTS void bitwise_and(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
\r
603 //! calculates per-element bit-wise "exclusive or" operation
\r
604 CV_EXPORTS void bitwise_xor(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, const GpuMat& mask=GpuMat(), Stream& stream = Stream::Null());
\r
606 //! computes per-element minimum of two arrays (dst = min(src1, src2))
\r
607 CV_EXPORTS void min(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
\r
609 //! computes per-element minimum of array and scalar (dst = min(src1, src2))
\r
610 CV_EXPORTS void min(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
\r
612 //! computes per-element maximum of two arrays (dst = max(src1, src2))
\r
613 CV_EXPORTS void max(const GpuMat& src1, const GpuMat& src2, GpuMat& dst, Stream& stream = Stream::Null());
\r
615 //! computes per-element maximum of array and scalar (dst = max(src1, src2))
\r
616 CV_EXPORTS void max(const GpuMat& src1, double src2, GpuMat& dst, Stream& stream = Stream::Null());
\r
618 //! computes the weighted sum of two arrays
\r
619 CV_EXPORTS void addWeighted(const GpuMat& src1, double alpha, const GpuMat& src2, double beta, double gamma, GpuMat& dst,
\r
620 int dtype = -1, Stream& stream = Stream::Null());
\r
623 ////////////////////////////// Image processing //////////////////////////////
\r
625 //! DST[x,y] = SRC[xmap[x,y],ymap[x,y]]
\r
626 //! supports only CV_32FC1 map type
\r
627 CV_EXPORTS void remap(const GpuMat& src, GpuMat& dst, const GpuMat& xmap, const GpuMat& ymap,
\r
628 int interpolation, int borderMode = BORDER_CONSTANT, const Scalar& borderValue = Scalar(),
\r
629 Stream& stream = Stream::Null());
\r
631 //! Does mean shift filtering on GPU.
\r
632 CV_EXPORTS void meanShiftFiltering(const GpuMat& src, GpuMat& dst, int sp, int sr,
\r
633 TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1));
\r
635 //! Does mean shift procedure on GPU.
\r
636 CV_EXPORTS void meanShiftProc(const GpuMat& src, GpuMat& dstr, GpuMat& dstsp, int sp, int sr,
\r
637 TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1));
\r
639 //! Does mean shift segmentation with elimination of small regions.
\r
640 CV_EXPORTS void meanShiftSegmentation(const GpuMat& src, Mat& dst, int sp, int sr, int minsize,
\r
641 TermCriteria criteria = TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 5, 1));
\r
643 //! Does coloring of disparity image: [0..ndisp) -> [0..240, 1, 1] in HSV.
\r
644 //! Supported types of input disparity: CV_8U, CV_16S.
\r
645 //! Output disparity has CV_8UC4 type in BGRA format (alpha = 255).
\r
646 CV_EXPORTS void drawColorDisp(const GpuMat& src_disp, GpuMat& dst_disp, int ndisp, Stream& stream = Stream::Null());
\r
648 //! Reprojects disparity image to 3D space.
\r
649 //! Supports CV_8U and CV_16S types of input disparity.
\r
650 //! The output is a 4-channel floating-point (CV_32FC4) matrix.
\r
651 //! Each element of this matrix will contain the 3D coordinates of the point (x,y,z,1), computed from the disparity map.
\r
652 //! Q is the 4x4 perspective transformation matrix that can be obtained with cvStereoRectify.
\r
653 CV_EXPORTS void reprojectImageTo3D(const GpuMat& disp, GpuMat& xyzw, const Mat& Q, Stream& stream = Stream::Null());
\r
655 //! converts image from one color space to another
\r
656 CV_EXPORTS void cvtColor(const GpuMat& src, GpuMat& dst, int code, int dcn = 0, Stream& stream = Stream::Null());
\r
658 //! applies fixed threshold to the image
\r
659 CV_EXPORTS double threshold(const GpuMat& src, GpuMat& dst, double thresh, double maxval, int type, Stream& stream = Stream::Null());
\r
661 //! resizes the image
\r
662 //! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
\r
663 CV_EXPORTS void resize(const GpuMat& src, GpuMat& dst, Size dsize, double fx=0, double fy=0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
\r
665 //! warps the image using affine transformation
\r
666 //! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
\r
667 CV_EXPORTS void warpAffine(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR, Stream& stream = Stream::Null());
\r
669 //! warps the image using perspective transformation
\r
670 //! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
\r
671 CV_EXPORTS void warpPerspective(const GpuMat& src, GpuMat& dst, const Mat& M, Size dsize, int flags = INTER_LINEAR, Stream& stream = Stream::Null());
\r
673 //! builds plane warping maps
\r
674 CV_EXPORTS void buildWarpPlaneMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, const Mat &T, float scale,
\r
675 GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
\r
677 //! builds cylindrical warping maps
\r
678 CV_EXPORTS void buildWarpCylindricalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
\r
679 GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
\r
681 //! builds spherical warping maps
\r
682 CV_EXPORTS void buildWarpSphericalMaps(Size src_size, Rect dst_roi, const Mat &K, const Mat& R, float scale,
\r
683 GpuMat& map_x, GpuMat& map_y, Stream& stream = Stream::Null());
\r
685 //! rotate 8bit single or four channel image
\r
686 //! Supports INTER_NEAREST, INTER_LINEAR, INTER_CUBIC
\r
687 //! supports CV_8UC1, CV_8UC4 types
\r
688 CV_EXPORTS void rotate(const GpuMat& src, GpuMat& dst, Size dsize, double angle, double xShift = 0, double yShift = 0, int interpolation = INTER_LINEAR, Stream& stream = Stream::Null());
\r
690 //! copies 2D array to a larger destination array and pads borders with user-specifiable constant
\r
691 CV_EXPORTS void copyMakeBorder(const GpuMat& src, GpuMat& dst, int top, int bottom, int left, int right, int borderType, const Scalar& value = Scalar(), Stream& stream = Stream::Null());
\r
693 //! computes the integral image
\r
694 //! sum will have CV_32S type, but will contain unsigned int values
\r
695 //! supports only CV_8UC1 source type
\r
696 CV_EXPORTS void integral(const GpuMat& src, GpuMat& sum, Stream& stream = Stream::Null());
\r
698 //! buffered version
\r
699 CV_EXPORTS void integralBuffered(const GpuMat& src, GpuMat& sum, GpuMat& buffer, Stream& stream = Stream::Null());
\r
701 //! computes the integral image and integral for the squared image
\r
702 //! sum will have CV_32S type, sqsum - CV32F type
\r
703 //! supports only CV_8UC1 source type
\r
704 CV_EXPORTS void integral(const GpuMat& src, GpuMat& sum, GpuMat& sqsum, Stream& stream = Stream::Null());
\r
706 //! computes squared integral image
\r
707 //! result matrix will have 64F type, but will contain 64U values
\r
708 //! supports source images of 8UC1 type only
\r
709 CV_EXPORTS void sqrIntegral(const GpuMat& src, GpuMat& sqsum, Stream& stream = Stream::Null());
\r
711 //! computes vertical sum, supports only CV_32FC1 images
\r
712 CV_EXPORTS void columnSum(const GpuMat& src, GpuMat& sum);
\r
714 //! computes the standard deviation of integral images
\r
715 //! supports only CV_32SC1 source type and CV_32FC1 sqr type
\r
716 //! output will have CV_32FC1 type
\r
717 CV_EXPORTS void rectStdDev(const GpuMat& src, const GpuMat& sqr, GpuMat& dst, const Rect& rect, Stream& stream = Stream::Null());
\r
719 //! computes Harris cornerness criteria at each image pixel
\r
720 CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, double k, int borderType=BORDER_REFLECT101);
\r
721 CV_EXPORTS void cornerHarris(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, double k, int borderType=BORDER_REFLECT101);
\r
723 //! computes minimum eigen value of 2x2 derivative covariation matrix at each pixel - the cornerness criteria
\r
724 CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
\r
725 CV_EXPORTS void cornerMinEigenVal(const GpuMat& src, GpuMat& dst, GpuMat& Dx, GpuMat& Dy, int blockSize, int ksize, int borderType=BORDER_REFLECT101);
\r
727 //! performs per-element multiplication of two full (not packed) Fourier spectrums
\r
728 //! supports 32FC2 matrixes only (interleaved format)
\r
729 CV_EXPORTS void mulSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags, bool conjB=false);
\r
731 //! performs per-element multiplication of two full (not packed) Fourier spectrums
\r
732 //! supports 32FC2 matrixes only (interleaved format)
\r
733 CV_EXPORTS void mulAndScaleSpectrums(const GpuMat& a, const GpuMat& b, GpuMat& c, int flags,
\r
734 float scale, bool conjB=false);
\r
736 //! Performs a forward or inverse discrete Fourier transform (1D or 2D) of floating point matrix.
\r
737 //! Param dft_size is the size of DFT transform.
\r
739 //! If the source matrix is not continous, then additional copy will be done,
\r
740 //! so to avoid copying ensure the source matrix is continous one. If you want to use
\r
741 //! preallocated output ensure it is continuous too, otherwise it will be reallocated.
\r
743 //! Being implemented via CUFFT real-to-complex transform result contains only non-redundant values
\r
744 //! in CUFFT's format. Result as full complex matrix for such kind of transform cannot be retrieved.
\r
746 //! For complex-to-real transform it is assumed that the source matrix is packed in CUFFT's format.
\r
747 CV_EXPORTS void dft(const GpuMat& src, GpuMat& dst, Size dft_size, int flags=0);
\r
749 //! computes convolution (or cross-correlation) of two images using discrete Fourier transform
\r
750 //! supports source images of 32FC1 type only
\r
751 //! result matrix will have 32FC1 type
\r
752 CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result,
\r
755 struct CV_EXPORTS ConvolveBuf;
\r
757 //! buffered version
\r
758 CV_EXPORTS void convolve(const GpuMat& image, const GpuMat& templ, GpuMat& result,
\r
759 bool ccorr, ConvolveBuf& buf);
\r
761 struct CV_EXPORTS ConvolveBuf
\r
764 ConvolveBuf(Size image_size, Size templ_size)
\r
765 { create(image_size, templ_size); }
\r
766 void create(Size image_size, Size templ_size);
\r
769 static Size estimateBlockSize(Size result_size, Size templ_size);
\r
770 friend void convolve(const GpuMat&, const GpuMat&, GpuMat&, bool, ConvolveBuf&);
\r
777 GpuMat image_spect, templ_spect, result_spect;
\r
778 GpuMat image_block, templ_block, result_data;
\r
781 //! computes the proximity map for the raster template and the image where the template is searched for
\r
782 CV_EXPORTS void matchTemplate(const GpuMat& image, const GpuMat& templ, GpuMat& result, int method);
\r
784 //! smoothes the source image and downsamples it
\r
785 CV_EXPORTS void pyrDown(const GpuMat& src, GpuMat& dst, int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
\r
787 //! upsamples the source image and then smoothes it
\r
788 CV_EXPORTS void pyrUp(const GpuMat& src, GpuMat& dst, int borderType = BORDER_DEFAULT, Stream& stream = Stream::Null());
\r
790 //! performs linear blending of two images
\r
791 //! to avoid accuracy errors sum of weigths shouldn't be very close to zero
\r
792 CV_EXPORTS void blendLinear(const GpuMat& img1, const GpuMat& img2, const GpuMat& weights1, const GpuMat& weights2,
\r
793 GpuMat& result, Stream& stream = Stream::Null());
\r
796 struct CV_EXPORTS CannyBuf;
\r
798 CV_EXPORTS void Canny(const GpuMat& image, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
\r
799 CV_EXPORTS void Canny(const GpuMat& image, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, int apperture_size = 3, bool L2gradient = false);
\r
800 CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
\r
801 CV_EXPORTS void Canny(const GpuMat& dx, const GpuMat& dy, CannyBuf& buf, GpuMat& edges, double low_thresh, double high_thresh, bool L2gradient = false);
\r
803 struct CV_EXPORTS CannyBuf
\r
806 explicit CannyBuf(const Size& image_size, int apperture_size = 3) {create(image_size, apperture_size);}
\r
807 CannyBuf(const GpuMat& dx_, const GpuMat& dy_);
\r
809 void create(const Size& image_size, int apperture_size = 3);
\r
814 GpuMat dx_buf, dy_buf;
\r
816 GpuMat trackBuf1, trackBuf2;
\r
817 Ptr<FilterEngine_GPU> filterDX, filterDY;
\r
820 ////////////////////////////// Matrix reductions //////////////////////////////
\r
822 //! computes mean value and standard deviation of all or selected array elements
\r
823 //! supports only CV_8UC1 type
\r
824 CV_EXPORTS void meanStdDev(const GpuMat& mtx, Scalar& mean, Scalar& stddev);
\r
826 //! computes norm of array
\r
827 //! supports NORM_INF, NORM_L1, NORM_L2
\r
828 //! supports all matrices except 64F
\r
829 CV_EXPORTS double norm(const GpuMat& src1, int normType=NORM_L2);
\r
831 //! computes norm of array
\r
832 //! supports NORM_INF, NORM_L1, NORM_L2
\r
833 //! supports all matrices except 64F
\r
834 CV_EXPORTS double norm(const GpuMat& src1, int normType, GpuMat& buf);
\r
836 //! computes norm of the difference between two arrays
\r
837 //! supports NORM_INF, NORM_L1, NORM_L2
\r
838 //! supports only CV_8UC1 type
\r
839 CV_EXPORTS double norm(const GpuMat& src1, const GpuMat& src2, int normType=NORM_L2);
\r
841 //! computes sum of array elements
\r
842 //! supports only single channel images
\r
843 CV_EXPORTS Scalar sum(const GpuMat& src);
\r
845 //! computes sum of array elements
\r
846 //! supports only single channel images
\r
847 CV_EXPORTS Scalar sum(const GpuMat& src, GpuMat& buf);
\r
849 //! computes sum of array elements absolute values
\r
850 //! supports only single channel images
\r
851 CV_EXPORTS Scalar absSum(const GpuMat& src);
\r
853 //! computes sum of array elements absolute values
\r
854 //! supports only single channel images
\r
855 CV_EXPORTS Scalar absSum(const GpuMat& src, GpuMat& buf);
\r
857 //! computes squared sum of array elements
\r
858 //! supports only single channel images
\r
859 CV_EXPORTS Scalar sqrSum(const GpuMat& src);
\r
861 //! computes squared sum of array elements
\r
862 //! supports only single channel images
\r
863 CV_EXPORTS Scalar sqrSum(const GpuMat& src, GpuMat& buf);
\r
865 //! finds global minimum and maximum array elements and returns their values
\r
866 CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal=0, const GpuMat& mask=GpuMat());
\r
868 //! finds global minimum and maximum array elements and returns their values
\r
869 CV_EXPORTS void minMax(const GpuMat& src, double* minVal, double* maxVal, const GpuMat& mask, GpuMat& buf);
\r
871 //! finds global minimum and maximum array elements and returns their values with locations
\r
872 CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal=0, Point* minLoc=0, Point* maxLoc=0,
\r
873 const GpuMat& mask=GpuMat());
\r
875 //! finds global minimum and maximum array elements and returns their values with locations
\r
876 CV_EXPORTS void minMaxLoc(const GpuMat& src, double* minVal, double* maxVal, Point* minLoc, Point* maxLoc,
\r
877 const GpuMat& mask, GpuMat& valbuf, GpuMat& locbuf);
\r
879 //! counts non-zero array elements
\r
880 CV_EXPORTS int countNonZero(const GpuMat& src);
\r
882 //! counts non-zero array elements
\r
883 CV_EXPORTS int countNonZero(const GpuMat& src, GpuMat& buf);
\r
885 //! reduces a matrix to a vector
\r
886 CV_EXPORTS void reduce(const GpuMat& mtx, GpuMat& vec, int dim, int reduceOp, int dtype = -1, Stream& stream = Stream::Null());
\r
889 ///////////////////////////// Calibration 3D //////////////////////////////////
\r
891 CV_EXPORTS void transformPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
\r
892 GpuMat& dst, Stream& stream = Stream::Null());
\r
894 CV_EXPORTS void projectPoints(const GpuMat& src, const Mat& rvec, const Mat& tvec,
\r
895 const Mat& camera_mat, const Mat& dist_coef, GpuMat& dst,
\r
896 Stream& stream = Stream::Null());
\r
898 CV_EXPORTS void solvePnPRansac(const Mat& object, const Mat& image, const Mat& camera_mat,
\r
899 const Mat& dist_coef, Mat& rvec, Mat& tvec, bool use_extrinsic_guess=false,
\r
900 int num_iters=100, float max_dist=8.0, int min_inlier_count=100,
\r
901 vector<int>* inliers=NULL);
\r
903 //////////////////////////////// Image Labeling ////////////////////////////////
\r
905 //!performs labeling via graph cuts
\r
906 CV_EXPORTS void graphcut(GpuMat& terminals, GpuMat& leftTransp, GpuMat& rightTransp, GpuMat& top, GpuMat& bottom, GpuMat& labels, GpuMat& buf, Stream& stream = Stream::Null());
\r
908 ////////////////////////////////// Histograms //////////////////////////////////
\r
910 //! Compute levels with even distribution. levels will have 1 row and nLevels cols and CV_32SC1 type.
\r
911 CV_EXPORTS void evenLevels(GpuMat& levels, int nLevels, int lowerLevel, int upperLevel);
\r
912 //! Calculates histogram with evenly distributed bins for signle channel source.
\r
913 //! Supports CV_8UC1, CV_16UC1 and CV_16SC1 source types.
\r
914 //! Output hist will have one row and histSize cols and CV_32SC1 type.
\r
915 CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
\r
916 CV_EXPORTS void histEven(const GpuMat& src, GpuMat& hist, GpuMat& buf, int histSize, int lowerLevel, int upperLevel, Stream& stream = Stream::Null());
\r
917 //! Calculates histogram with evenly distributed bins for four-channel source.
\r
918 //! All channels of source are processed separately.
\r
919 //! Supports CV_8UC4, CV_16UC4 and CV_16SC4 source types.
\r
920 //! Output hist[i] will have one row and histSize[i] cols and CV_32SC1 type.
\r
921 CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
\r
922 CV_EXPORTS void histEven(const GpuMat& src, GpuMat hist[4], GpuMat& buf, int histSize[4], int lowerLevel[4], int upperLevel[4], Stream& stream = Stream::Null());
\r
923 //! Calculates histogram with bins determined by levels array.
\r
924 //! levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
\r
925 //! Supports CV_8UC1, CV_16UC1, CV_16SC1 and CV_32FC1 source types.
\r
926 //! Output hist will have one row and (levels.cols-1) cols and CV_32SC1 type.
\r
927 CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, Stream& stream = Stream::Null());
\r
928 CV_EXPORTS void histRange(const GpuMat& src, GpuMat& hist, const GpuMat& levels, GpuMat& buf, Stream& stream = Stream::Null());
\r
929 //! Calculates histogram with bins determined by levels array.
\r
930 //! All levels must have one row and CV_32SC1 type if source has integer type or CV_32FC1 otherwise.
\r
931 //! All channels of source are processed separately.
\r
932 //! Supports CV_8UC4, CV_16UC4, CV_16SC4 and CV_32FC4 source types.
\r
933 //! Output hist[i] will have one row and (levels[i].cols-1) cols and CV_32SC1 type.
\r
934 CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], Stream& stream = Stream::Null());
\r
935 CV_EXPORTS void histRange(const GpuMat& src, GpuMat hist[4], const GpuMat levels[4], GpuMat& buf, Stream& stream = Stream::Null());
\r
937 //! Calculates histogram for 8u one channel image
\r
938 //! Output hist will have one row, 256 cols and CV32SC1 type.
\r
939 CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, Stream& stream = Stream::Null());
\r
940 CV_EXPORTS void calcHist(const GpuMat& src, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
\r
942 //! normalizes the grayscale image brightness and contrast by normalizing its histogram
\r
943 CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, Stream& stream = Stream::Null());
\r
944 CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, Stream& stream = Stream::Null());
\r
945 CV_EXPORTS void equalizeHist(const GpuMat& src, GpuMat& dst, GpuMat& hist, GpuMat& buf, Stream& stream = Stream::Null());
\r
947 //////////////////////////////// StereoBM_GPU ////////////////////////////////
\r
949 class CV_EXPORTS StereoBM_GPU
\r
952 enum { BASIC_PRESET = 0, PREFILTER_XSOBEL = 1 };
\r
954 enum { DEFAULT_NDISP = 64, DEFAULT_WINSZ = 19 };
\r
956 //! the default constructor
\r
958 //! the full constructor taking the camera-specific preset, number of disparities and the SAD window size. ndisparities must be multiple of 8.
\r
959 StereoBM_GPU(int preset, int ndisparities = DEFAULT_NDISP, int winSize = DEFAULT_WINSZ);
\r
961 //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair
\r
962 //! Output disparity has CV_8U type.
\r
963 void operator() ( const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
\r
965 //! Some heuristics that tries to estmate
\r
966 // if current GPU will be faster than CPU in this algorithm.
\r
967 // It queries current active device.
\r
968 static bool checkIfGpuCallReasonable();
\r
974 // If avergeTexThreshold == 0 => post procesing is disabled
\r
975 // If avergeTexThreshold != 0 then disparity is set 0 in each point (x,y) where for left image
\r
976 // SumOfHorizontalGradiensInWindow(x, y, winSize) < (winSize * winSize) * avergeTexThreshold
\r
977 // i.e. input left image is low textured.
\r
978 float avergeTexThreshold;
\r
980 GpuMat minSSD, leBuf, riBuf;
\r
983 ////////////////////////// StereoBeliefPropagation ///////////////////////////
\r
984 // "Efficient Belief Propagation for Early Vision"
\r
987 class CV_EXPORTS StereoBeliefPropagation
\r
990 enum { DEFAULT_NDISP = 64 };
\r
991 enum { DEFAULT_ITERS = 5 };
\r
992 enum { DEFAULT_LEVELS = 5 };
\r
994 static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels);
\r
996 //! the default constructor
\r
997 explicit StereoBeliefPropagation(int ndisp = DEFAULT_NDISP,
\r
998 int iters = DEFAULT_ITERS,
\r
999 int levels = DEFAULT_LEVELS,
\r
1000 int msg_type = CV_32F);
\r
1002 //! the full constructor taking the number of disparities, number of BP iterations on each level,
\r
1003 //! number of levels, truncation of data cost, data weight,
\r
1004 //! truncation of discontinuity cost and discontinuity single jump
\r
1005 //! DataTerm = data_weight * min(fabs(I2-I1), max_data_term)
\r
1006 //! DiscTerm = min(disc_single_jump * fabs(f1-f2), max_disc_term)
\r
1007 //! please see paper for more details
\r
1008 StereoBeliefPropagation(int ndisp, int iters, int levels,
\r
1009 float max_data_term, float data_weight,
\r
1010 float max_disc_term, float disc_single_jump,
\r
1011 int msg_type = CV_32F);
\r
1013 //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
\r
1014 //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
\r
1015 void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
\r
1018 //! version for user specified data term
\r
1019 void operator()(const GpuMat& data, GpuMat& disparity, Stream& stream = Stream::Null());
\r
1026 float max_data_term;
\r
1027 float data_weight;
\r
1028 float max_disc_term;
\r
1029 float disc_single_jump;
\r
1033 GpuMat u, d, l, r, u2, d2, l2, r2;
\r
1034 std::vector<GpuMat> datas;
\r
1038 /////////////////////////// StereoConstantSpaceBP ///////////////////////////
\r
1039 // "A Constant-Space Belief Propagation Algorithm for Stereo Matching"
\r
1040 // Qingxiong Yang, Liang Wang, Narendra Ahuja
\r
1041 // http://vision.ai.uiuc.edu/~qyang6/
\r
1043 class CV_EXPORTS StereoConstantSpaceBP
\r
1046 enum { DEFAULT_NDISP = 128 };
\r
1047 enum { DEFAULT_ITERS = 8 };
\r
1048 enum { DEFAULT_LEVELS = 4 };
\r
1049 enum { DEFAULT_NR_PLANE = 4 };
\r
1051 static void estimateRecommendedParams(int width, int height, int& ndisp, int& iters, int& levels, int& nr_plane);
\r
1053 //! the default constructor
\r
1054 explicit StereoConstantSpaceBP(int ndisp = DEFAULT_NDISP,
\r
1055 int iters = DEFAULT_ITERS,
\r
1056 int levels = DEFAULT_LEVELS,
\r
1057 int nr_plane = DEFAULT_NR_PLANE,
\r
1058 int msg_type = CV_32F);
\r
1060 //! the full constructor taking the number of disparities, number of BP iterations on each level,
\r
1061 //! number of levels, number of active disparity on the first level, truncation of data cost, data weight,
\r
1062 //! truncation of discontinuity cost, discontinuity single jump and minimum disparity threshold
\r
1063 StereoConstantSpaceBP(int ndisp, int iters, int levels, int nr_plane,
\r
1064 float max_data_term, float data_weight, float max_disc_term, float disc_single_jump,
\r
1065 int min_disp_th = 0,
\r
1066 int msg_type = CV_32F);
\r
1068 //! the stereo correspondence operator. Finds the disparity for the specified rectified stereo pair,
\r
1069 //! if disparity is empty output type will be CV_16S else output type will be disparity.type().
\r
1070 void operator()(const GpuMat& left, const GpuMat& right, GpuMat& disparity, Stream& stream = Stream::Null());
\r
1079 float max_data_term;
\r
1080 float data_weight;
\r
1081 float max_disc_term;
\r
1082 float disc_single_jump;
\r
1088 bool use_local_init_data_cost;
\r
1090 GpuMat u[2], d[2], l[2], r[2];
\r
1091 GpuMat disp_selected_pyr[2];
\r
1094 GpuMat data_cost_selected;
\r
1101 /////////////////////////// DisparityBilateralFilter ///////////////////////////
\r
1102 // Disparity map refinement using joint bilateral filtering given a single color image.
\r
1103 // Qingxiong Yang, Liang Wang, Narendra Ahuja
\r
1104 // http://vision.ai.uiuc.edu/~qyang6/
\r
1106 class CV_EXPORTS DisparityBilateralFilter
\r
1109 enum { DEFAULT_NDISP = 64 };
\r
1110 enum { DEFAULT_RADIUS = 3 };
\r
1111 enum { DEFAULT_ITERS = 1 };
\r
1113 //! the default constructor
\r
1114 explicit DisparityBilateralFilter(int ndisp = DEFAULT_NDISP, int radius = DEFAULT_RADIUS, int iters = DEFAULT_ITERS);
\r
1116 //! the full constructor taking the number of disparities, filter radius,
\r
1117 //! number of iterations, truncation of data continuity, truncation of disparity continuity
\r
1118 //! and filter range sigma
\r
1119 DisparityBilateralFilter(int ndisp, int radius, int iters, float edge_threshold, float max_disc_threshold, float sigma_range);
\r
1121 //! the disparity map refinement operator. Refine disparity map using joint bilateral filtering given a single color image.
\r
1122 //! disparity must have CV_8U or CV_16S type, image must have CV_8UC1 or CV_8UC3 type.
\r
1123 void operator()(const GpuMat& disparity, const GpuMat& image, GpuMat& dst, Stream& stream = Stream::Null());
\r
1130 float edge_threshold;
\r
1131 float max_disc_threshold;
\r
1132 float sigma_range;
\r
1134 GpuMat table_color;
\r
1135 GpuMat table_space;
\r
1139 //////////////// HOG (Histogram-of-Oriented-Gradients) Descriptor and Object Detector //////////////
\r
1141 struct CV_EXPORTS HOGDescriptor
\r
1143 enum { DEFAULT_WIN_SIGMA = -1 };
\r
1144 enum { DEFAULT_NLEVELS = 64 };
\r
1145 enum { DESCR_FORMAT_ROW_BY_ROW, DESCR_FORMAT_COL_BY_COL };
\r
1147 HOGDescriptor(Size win_size=Size(64, 128), Size block_size=Size(16, 16),
\r
1148 Size block_stride=Size(8, 8), Size cell_size=Size(8, 8),
\r
1149 int nbins=9, double win_sigma=DEFAULT_WIN_SIGMA,
\r
1150 double threshold_L2hys=0.2, bool gamma_correction=true,
\r
1151 int nlevels=DEFAULT_NLEVELS);
\r
1153 size_t getDescriptorSize() const;
\r
1154 size_t getBlockHistogramSize() const;
\r
1156 void setSVMDetector(const vector<float>& detector);
\r
1158 static vector<float> getDefaultPeopleDetector();
\r
1159 static vector<float> getPeopleDetector48x96();
\r
1160 static vector<float> getPeopleDetector64x128();
\r
1162 void detect(const GpuMat& img, vector<Point>& found_locations,
\r
1163 double hit_threshold=0, Size win_stride=Size(),
\r
1164 Size padding=Size());
\r
1166 void detectMultiScale(const GpuMat& img, vector<Rect>& found_locations,
\r
1167 double hit_threshold=0, Size win_stride=Size(),
\r
1168 Size padding=Size(), double scale0=1.05,
\r
1169 int group_threshold=2);
\r
1171 void getDescriptors(const GpuMat& img, Size win_stride,
\r
1172 GpuMat& descriptors,
\r
1173 int descr_format=DESCR_FORMAT_COL_BY_COL);
\r
1177 Size block_stride;
\r
1181 double threshold_L2hys;
\r
1182 bool gamma_correction;
\r
1186 void computeBlockHistograms(const GpuMat& img);
\r
1187 void computeGradient(const GpuMat& img, GpuMat& grad, GpuMat& qangle);
\r
1189 double getWinSigma() const;
\r
1190 bool checkDetectorSize() const;
\r
1192 static int numPartsWithin(int size, int part_size, int stride);
\r
1193 static Size numPartsWithin(Size size, Size part_size, Size stride);
\r
1195 // Coefficients of the separating plane
\r
1199 // Results of the last classification step
\r
1200 GpuMat labels, labels_buf;
\r
1203 // Results of the last histogram evaluation step
\r
1204 GpuMat block_hists, block_hists_buf;
\r
1206 // Gradients conputation results
\r
1207 GpuMat grad, qangle, grad_buf, qangle_buf;
\r
1209 // returns subbuffer with required size, reallocates buffer if nessesary.
\r
1210 static GpuMat getBuffer(const Size& sz, int type, GpuMat& buf);
\r
1211 static GpuMat getBuffer(int rows, int cols, int type, GpuMat& buf);
\r
1213 std::vector<GpuMat> image_scales;
\r
1217 ////////////////////////////////// BruteForceMatcher //////////////////////////////////
\r
1219 class CV_EXPORTS BruteForceMatcher_GPU_base
\r
1222 enum DistType {L1Dist = 0, L2Dist, HammingDist};
\r
1224 explicit BruteForceMatcher_GPU_base(DistType distType = L2Dist);
\r
1226 // Add descriptors to train descriptor collection.
\r
1227 void add(const std::vector<GpuMat>& descCollection);
\r
1229 // Get train descriptors collection.
\r
1230 const std::vector<GpuMat>& getTrainDescriptors() const;
\r
1232 // Clear train descriptors collection.
\r
1235 // Return true if there are not train descriptors in collection.
\r
1236 bool empty() const;
\r
1238 // Return true if the matcher supports mask in match methods.
\r
1239 bool isMaskSupported() const;
\r
1241 // Find one best match for each query descriptor.
\r
1242 // trainIdx.at<int>(0, queryIdx) will contain best train index for queryIdx
\r
1243 // distance.at<float>(0, queryIdx) will contain distance
\r
1244 void matchSingle(const GpuMat& queryDescs, const GpuMat& trainDescs,
\r
1245 GpuMat& trainIdx, GpuMat& distance,
\r
1246 const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
\r
1248 // Download trainIdx and distance and convert it to CPU vector with DMatch
\r
1249 static void matchDownload(const GpuMat& trainIdx, const GpuMat& distance, std::vector<DMatch>& matches);
\r
1250 // Convert trainIdx and distance to vector with DMatch
\r
1251 static void matchConvert(const Mat& trainIdx, const Mat& distance, std::vector<DMatch>& matches);
\r
1253 // Find one best match for each query descriptor.
\r
1254 void match(const GpuMat& queryDescs, const GpuMat& trainDescs, std::vector<DMatch>& matches,
\r
1255 const GpuMat& mask = GpuMat());
\r
1257 // Make gpu collection of trains and masks in suitable format for matchCollection function
\r
1258 void makeGpuCollection(GpuMat& trainCollection, GpuMat& maskCollection,
\r
1259 const vector<GpuMat>& masks = std::vector<GpuMat>());
\r
1261 // Find one best match from train collection for each query descriptor.
\r
1262 // trainIdx.at<int>(0, queryIdx) will contain best train index for queryIdx
\r
1263 // imgIdx.at<int>(0, queryIdx) will contain best image index for queryIdx
\r
1264 // distance.at<float>(0, queryIdx) will contain distance
\r
1265 void matchCollection(const GpuMat& queryDescs, const GpuMat& trainCollection,
\r
1266 GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance,
\r
1267 const GpuMat& maskCollection, Stream& stream = Stream::Null());
\r
1269 // Download trainIdx, imgIdx and distance and convert it to vector with DMatch
\r
1270 static void matchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, std::vector<DMatch>& matches);
\r
1271 // Convert trainIdx, imgIdx and distance to vector with DMatch
\r
1272 static void matchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, std::vector<DMatch>& matches);
\r
1274 // Find one best match from train collection for each query descriptor.
\r
1275 void match(const GpuMat& queryDescs, std::vector<DMatch>& matches, const std::vector<GpuMat>& masks = std::vector<GpuMat>());
\r
1277 // Find k best matches for each query descriptor (in increasing order of distances).
\r
1278 // trainIdx.at<int>(queryIdx, i) will contain index of i'th best trains (i < k).
\r
1279 // distance.at<float>(queryIdx, i) will contain distance.
\r
1280 // allDist is a buffer to store all distance between query descriptors and train descriptors
\r
1281 // it have size (nQuery,nTrain) and CV_32F type
\r
1282 // allDist.at<float>(queryIdx, trainIdx) will contain FLT_MAX, if trainIdx is one from k best,
\r
1283 // otherwise it will contain distance between queryIdx and trainIdx descriptors
\r
1284 void knnMatch(const GpuMat& queryDescs, const GpuMat& trainDescs,
\r
1285 GpuMat& trainIdx, GpuMat& distance, GpuMat& allDist, int k, const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
\r
1287 // Download trainIdx and distance and convert it to vector with DMatch
\r
1288 // compactResult is used when mask is not empty. If compactResult is false matches
\r
1289 // vector will have the same size as queryDescriptors rows. If compactResult is true
\r
1290 // matches vector will not contain matches for fully masked out query descriptors.
\r
1291 static void knnMatchDownload(const GpuMat& trainIdx, const GpuMat& distance,
\r
1292 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1293 // Convert trainIdx and distance to vector with DMatch
\r
1294 static void knnMatchConvert(const Mat& trainIdx, const Mat& distance,
\r
1295 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1297 // Find k best matches for each query descriptor (in increasing order of distances).
\r
1298 // compactResult is used when mask is not empty. If compactResult is false matches
\r
1299 // vector will have the same size as queryDescriptors rows. If compactResult is true
\r
1300 // matches vector will not contain matches for fully masked out query descriptors.
\r
1301 void knnMatch(const GpuMat& queryDescs, const GpuMat& trainDescs,
\r
1302 std::vector< std::vector<DMatch> >& matches, int k, const GpuMat& mask = GpuMat(),
\r
1303 bool compactResult = false);
\r
1305 // Find k best matches for each query descriptor (in increasing order of distances).
\r
1306 // compactResult is used when mask is not empty. If compactResult is false matches
\r
1307 // vector will have the same size as queryDescriptors rows. If compactResult is true
\r
1308 // matches vector will not contain matches for fully masked out query descriptors.
\r
1309 void knnMatch(const GpuMat& queryDescs, std::vector< std::vector<DMatch> >& matches, int knn,
\r
1310 const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false );
\r
1312 // Find best matches for each query descriptor which have distance less than maxDistance.
\r
1313 // nMatches.at<int>(0, queryIdx) will contain matches count for queryIdx.
\r
1314 // carefully nMatches can be greater than trainIdx.cols - it means that matcher didn't find all matches,
\r
1315 // because it didn't have enough memory.
\r
1316 // trainIdx.at<int>(queruIdx, i) will contain ith train index (i < min(nMatches.at<int>(0, queruIdx), trainIdx.cols))
\r
1317 // distance.at<int>(queruIdx, i) will contain ith distance (i < min(nMatches.at<int>(0, queruIdx), trainIdx.cols))
\r
1318 // If trainIdx is empty, then trainIdx and distance will be created with size nQuery x (nTrain / 2),
\r
1319 // otherwize user can pass own allocated trainIdx and distance with size nQuery x nMaxMatches
\r
1320 // Matches doesn't sorted.
\r
1321 void radiusMatchSingle(const GpuMat& queryDescs, const GpuMat& trainDescs,
\r
1322 GpuMat& trainIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
\r
1323 const GpuMat& mask = GpuMat(), Stream& stream = Stream::Null());
\r
1325 // Download trainIdx, nMatches and distance and convert it to vector with DMatch.
\r
1326 // matches will be sorted in increasing order of distances.
\r
1327 // compactResult is used when mask is not empty. If compactResult is false matches
\r
1328 // vector will have the same size as queryDescriptors rows. If compactResult is true
\r
1329 // matches vector will not contain matches for fully masked out query descriptors.
\r
1330 static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& distance, const GpuMat& nMatches,
\r
1331 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1332 // Convert trainIdx, nMatches and distance to vector with DMatch.
\r
1333 static void radiusMatchConvert(const Mat& trainIdx, const Mat& distance, const Mat& nMatches,
\r
1334 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1336 // Find best matches for each query descriptor which have distance less than maxDistance
\r
1337 // in increasing order of distances).
\r
1338 void radiusMatch(const GpuMat& queryDescs, const GpuMat& trainDescs,
\r
1339 std::vector< std::vector<DMatch> >& matches, float maxDistance,
\r
1340 const GpuMat& mask = GpuMat(), bool compactResult = false);
\r
1342 // Find best matches for each query descriptor which have distance less than maxDistance.
\r
1343 // Matches doesn't sorted.
\r
1344 void radiusMatchCollection(const GpuMat& queryDescs, const GpuMat& trainCollection,
\r
1345 GpuMat& trainIdx, GpuMat& imgIdx, GpuMat& distance, GpuMat& nMatches, float maxDistance,
\r
1346 const GpuMat& maskCollection, Stream& stream = Stream::Null());
\r
1348 // Download trainIdx, imgIdx, nMatches and distance and convert it to vector with DMatch.
\r
1349 // matches will be sorted in increasing order of distances.
\r
1350 // compactResult is used when mask is not empty. If compactResult is false matches
\r
1351 // vector will have the same size as queryDescriptors rows. If compactResult is true
\r
1352 // matches vector will not contain matches for fully masked out query descriptors.
\r
1353 static void radiusMatchDownload(const GpuMat& trainIdx, const GpuMat& imgIdx, const GpuMat& distance, const GpuMat& nMatches,
\r
1354 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1355 // Convert trainIdx, nMatches and distance to vector with DMatch.
\r
1356 static void radiusMatchConvert(const Mat& trainIdx, const Mat& imgIdx, const Mat& distance, const Mat& nMatches,
\r
1357 std::vector< std::vector<DMatch> >& matches, bool compactResult = false);
\r
1359 // Find best matches from train collection for each query descriptor which have distance less than
\r
1360 // maxDistance (in increasing order of distances).
\r
1361 void radiusMatch(const GpuMat& queryDescs, std::vector< std::vector<DMatch> >& matches, float maxDistance,
\r
1362 const std::vector<GpuMat>& masks = std::vector<GpuMat>(), bool compactResult = false);
\r
1364 DistType distType;
\r
1367 std::vector<GpuMat> trainDescCollection;
\r
1370 template <class Distance>
\r
1371 class CV_EXPORTS BruteForceMatcher_GPU;
\r
1373 template <typename T>
\r
1374 class CV_EXPORTS BruteForceMatcher_GPU< L1<T> > : public BruteForceMatcher_GPU_base
\r
1377 explicit BruteForceMatcher_GPU() : BruteForceMatcher_GPU_base(L1Dist) {}
\r
1378 explicit BruteForceMatcher_GPU(L1<T> /*d*/) : BruteForceMatcher_GPU_base(L1Dist) {}
\r
1380 template <typename T>
\r
1381 class CV_EXPORTS BruteForceMatcher_GPU< L2<T> > : public BruteForceMatcher_GPU_base
\r
1384 explicit BruteForceMatcher_GPU() : BruteForceMatcher_GPU_base(L2Dist) {}
\r
1385 explicit BruteForceMatcher_GPU(L2<T> /*d*/) : BruteForceMatcher_GPU_base(L2Dist) {}
\r
1387 template <> class CV_EXPORTS BruteForceMatcher_GPU< HammingLUT > : public BruteForceMatcher_GPU_base
\r
1390 explicit BruteForceMatcher_GPU() : BruteForceMatcher_GPU_base(HammingDist) {}
\r
1391 explicit BruteForceMatcher_GPU(HammingLUT /*d*/) : BruteForceMatcher_GPU_base(HammingDist) {}
\r
1393 template <> class CV_EXPORTS BruteForceMatcher_GPU< Hamming > : public BruteForceMatcher_GPU_base
\r
1396 explicit BruteForceMatcher_GPU() : BruteForceMatcher_GPU_base(HammingDist) {}
\r
1397 explicit BruteForceMatcher_GPU(Hamming /*d*/) : BruteForceMatcher_GPU_base(HammingDist) {}
\r
1400 ////////////////////////////////// CascadeClassifier_GPU //////////////////////////////////////////
\r
1401 // The cascade classifier class for object detection.
\r
1402 class CV_EXPORTS CascadeClassifier_GPU
\r
1405 CascadeClassifier_GPU();
\r
1406 CascadeClassifier_GPU(const string& filename);
\r
1407 ~CascadeClassifier_GPU();
\r
1409 bool empty() const;
\r
1410 bool load(const string& filename);
\r
1413 /* returns number of detected objects */
\r
1414 int detectMultiScale( const GpuMat& image, GpuMat& objectsBuf, double scaleFactor=1.2, int minNeighbors=4, Size minSize=Size());
\r
1416 bool findLargestObject;
\r
1417 bool visualizeInPlace;
\r
1419 Size getClassifierSize() const;
\r
1422 struct CascadeClassifierImpl;
\r
1423 CascadeClassifierImpl* impl;
\r
1426 ////////////////////////////////// SURF //////////////////////////////////////////
\r
1428 class CV_EXPORTS SURF_GPU : public CvSURFParams
\r
1431 enum KeypointLayout
\r
1442 //! the default constructor
\r
1444 //! the full constructor taking all the necessary parameters
\r
1445 explicit SURF_GPU(double _hessianThreshold, int _nOctaves=4,
\r
1446 int _nOctaveLayers=2, bool _extended=false, float _keypointsRatio=0.01f, bool _upright = false);
\r
1448 //! returns the descriptor size in float's (64 or 128)
\r
1449 int descriptorSize() const;
\r
1451 //! upload host keypoints to device memory
\r
1452 void uploadKeypoints(const vector<KeyPoint>& keypoints, GpuMat& keypointsGPU);
\r
1453 //! download keypoints from device to host memory
\r
1454 void downloadKeypoints(const GpuMat& keypointsGPU, vector<KeyPoint>& keypoints);
\r
1456 //! download descriptors from device to host memory
\r
1457 void downloadDescriptors(const GpuMat& descriptorsGPU, vector<float>& descriptors);
\r
1459 //! finds the keypoints using fast hessian detector used in SURF
\r
1460 //! supports CV_8UC1 images
\r
1461 //! keypoints will have nFeature cols and 6 rows
\r
1462 //! keypoints.ptr<float>(SF_X)[i] will contain x coordinate of i'th feature
\r
1463 //! keypoints.ptr<float>(SF_Y)[i] will contain y coordinate of i'th feature
\r
1464 //! keypoints.ptr<float>(SF_LAPLACIAN)[i] will contain laplacian sign of i'th feature
\r
1465 //! keypoints.ptr<float>(SF_SIZE)[i] will contain size of i'th feature
\r
1466 //! keypoints.ptr<float>(SF_DIR)[i] will contain orientation of i'th feature
\r
1467 //! keypoints.ptr<float>(SF_HESSIAN)[i] will contain response of i'th feature
\r
1468 void operator()(const GpuMat& img, const GpuMat& mask, GpuMat& keypoints);
\r
1469 //! finds the keypoints and computes their descriptors.
\r
1470 //! Optionally it can compute descriptors for the user-provided keypoints and recompute keypoints direction
\r
1471 void operator()(const GpuMat& img, const GpuMat& mask, GpuMat& keypoints, GpuMat& descriptors,
\r
1472 bool useProvidedKeypoints = false);
\r
1474 void operator()(const GpuMat& img, const GpuMat& mask, std::vector<KeyPoint>& keypoints);
\r
1475 void operator()(const GpuMat& img, const GpuMat& mask, std::vector<KeyPoint>& keypoints, GpuMat& descriptors,
\r
1476 bool useProvidedKeypoints = false);
\r
1478 void operator()(const GpuMat& img, const GpuMat& mask, std::vector<KeyPoint>& keypoints, std::vector<float>& descriptors,
\r
1479 bool useProvidedKeypoints = false);
\r
1481 void releaseMemory();
\r
1483 //! max keypoints = min(keypointsRatio * img.size().area(), 65535)
\r
1484 float keypointsRatio;
\r
1486 GpuMat sum, mask1, maskSum, intBuffer;
\r
1488 GpuMat det, trace;
\r
1490 GpuMat maxPosBuffer;
\r
1495 //! Speckle filtering - filters small connected components on diparity image.
\r
1496 //! It sets pixel (x,y) to newVal if it coresponds to small CC with size < maxSpeckleSize.
\r
1497 //! Threshold for border between CC is diffThreshold;
\r
1498 CV_EXPORTS void filterSpeckles( Mat& img, uchar newVal, int maxSpeckleSize, uchar diffThreshold, Mat& buf);
\r
1501 #include "opencv2/gpu/matrix_operations.hpp"
\r
1503 #endif /* __OPENCV_GPU_HPP__ */
\r