OpenCV - OpenCV DevZone

2

3

{{>toc}}

4

1

5

1

h2. 2.1

6

1

7

1

_*(April, 2010)*_

8

1

9

1

h3. General Modifications

10

1

11

1

  * SVN repository has been migrated from Source Forge to https://code.ros.org/svn/opencv.

12

1

    The bug tracker has been moved to https://code.ros.org/trac/opencv/.

13

1

    And we also have OpenCV twitter! http://twitter.com/opencvlibrary

14

1

15

1

  * The whole OpenCV is now using exceptions instead of the old libc-style mechanism.

16

1

    That is, instead of checking error code with cvGetErrStatus() (which currently always returns 0)

17

1

    you can now just call OpenCV functions inside C++ try-catch statements,

18

1

    cv::Exception is now derived from std::exception.

19

1

20

1

  * OpenCV does not support autotools-based build scripts,

21

1

    CMake (www.cmake.org) is the only way to build OpenCV on any OS.

22

1

    See http://opencv.willowgarage.com/wiki/InstallGuide.

23

1

24

1

  * All the parallel loops in OpenCV have been converted from OpenMP

25

1

    to Intel TBB (http://www.threadingbuildingblocks.org/). Thus parallel version of OpenCV

26

1

    can now be built using MSVC 2008 Express Edition or using earlier than 4.2 versions of GCC.

27

1

28

1

  * SWIG-based Python wrappers are still included,

29

1

    but they are not built by default and it's generally preferable to use the new wrappers.

30

1

    The python samples have been rewritten by James Bowman to use the new-style Python wrappers,

31

1

    which have been also created by James.

32

1

33

1

  * OpenCV can now be built and run in 64-bit mode on MacOSX 10.6 and Windows (see HighGUI and known problems below).

34

1

    On Windows both MSVC 2008 and mingw64 are known to work.

35

1

36

1

  * In theory OpenCV is now able to determine the host CPU on-fly and make use of SSE/SSE2/... instructions,

37

1

    if they are available. That is, it should be more safe to use WITH_SSE* flags in CMake.

38

1

    However, if you want maximum portability, it's recommended to turn on just WITH_SSE and WITH_SSE2

39

1

    and leave other SSE* turned off, as we found that using WITH_SSE3, WITH_SSSE3 and WITH_SSE4_1 can yield

40

1

    the code incompatible with Intel's pre-Penryn or AMD chips.

41

1

42

1

  * Experimental "static" OpenCV configuration in CMake was contributed by Jose Luis Blanco.

43

1

    Pass "BUILD_SHARED_LIBS=OFF" to CMake to build OpenCV statically.

44

1

45

1

h3. New Functionality

46

1

47

1

  # cxcore, cv, cvaux:

48

1

49

1

    * Grabcut (http://en.wikipedia.org/wiki/GrabCut) image segmentation algorithm has been implemented.

50

1

      See opencv/samples/c/grabcut.cpp

51

1

    * new improved version of one-way descriptor is added. See opencv/samples/c/one_way_sample.cpp

52

1

    * modified version of H. Hirschmuller semi-global stereo matching algorithm that we call SGBM

53

1

      (semi-global block matching) has been created. It is much faster than Kolmogorov's graph

54

1

      cuts-based algorithm and yet it's usually better than the block matching StereoBM algorithm.

55

1

      See opencv/samples/c/stereo_matching.cpp.

56

1

    * existing StereoBM stereo correspondence algorithm by K. Konolige was noticeably improved:

57

1

      added the optional left-right consistency check and speckle filtering,

58

1

      improved performance (by ~20%).

59

1

    * User can now control the image areas visible after the stereo rectification

60

1

      (see the extended stereoRectify/cvStereoRectify), and also limit the region

61

1

      where the disparity is computed (see CvStereoBMState::roi1, roi2; getValidDisparityROI).

62

1

    * Mixture-of-Gaussian based background subtraction algorithm has been rewritten for better performance

63

1

      and better accuracy. Alternative C++ interface Background Subtractor has been provided,

64

1

      along with the possibility to use the trained background model to segment the foreground

65

1

      without updating the model. See opencv/samples/c/bgfg_segm.cpp.

66

1

  # highgui:

67

1

68

1

    * MacOSX: OpenCV now includes Cocoa and QTKit backends, in addition to Carbon and Quicktime.

69

1

      Therefore you can build OpenCV as 64-bit library. Thanks to Andre Cohen and Nicolas Butko, which components.

70

1

      Note however that the backend are now in the alpha state, they can crash or leak memory,

71

1

      so for anything more serious than quick experiments you may prefer to use Carbon and Quicktime.

72

1

      To do that, pass USE_CARBON=ON and USE_QUICKTIME=ON to CMake and build OpenCV in 32-bit mode

73

1

      (i.e. select i386 architecture in Xcode).

74

1

    * Windows. OpenCV can now be built in 64-bit mode with MSVC 2008 and also mingw64.

75

1

    * Fullscreen has been added (thanks to Yannick Verdie).

76

1

      Call cvSetWindowProperty(window_name, CV_WINDOW_FULLSCREEN, 1) to make the particular window

77

1

      to fill the whole screen. This feature is not supported in the Cocoa bindings yet.

78

1

    * gstreamer backend has been improved a lot (thanks to Stefano Fabri)

79

1

80

1

h3. New Tests

81

1

82

1

  * A few dozens of new tests have been written and many existing tests have been extended

83

1

    to verify OpenCV correctness thoroughly. As a result, we brought the test coverage from

84

1

    rather mediocre numbers to pretty impressive ones (especially for cxcore and cv)!

85

1

86

1

    | Module  |    OpenCV 2.0 coverage, % (functions/conditions) |  OpenCV 2.1 coverage, % (functions/conditions) |

87

1

    | cxcore  |                     65/54                     |       91/73                        |

88

1

    | cv      |                     52/46                     |       80/68                        |

89

1

    | ml      |                     66/47                     |       73/52                        |

90

1

    | highgui |                     17/3                      |       54/27                        |

91

1

    | cvaux   |                     0/0                       |       5/12                         |

92

1

93

1

  * Many new regression tests have been written in Python that check both OpenCV and the new-style bindings.

94

1

95

1

  * The test data moved to the separate repository: https://code.ros.org/svn/opencv/trunk/opencv_extra/testdata.

96

1

    And it is not included into the package, thus some tests from the cvtest and mltest will report about the missing data.

97

1

    You can download the directory to your hard drive and run cvtest like:

98

1

    <pre><code>    ./cvtest -d <path_to_opencv_extra>/testdata/cv

99

1

    ./mltest -d <path_to_opencv_extra>/testdata/ml

100

1

    </code></pre>

101

1

102

1

  * The test engine has been improved:

103

1

     added flags -tn, -seed, -r

104

1

     the detailed information about failed tests is displayed right in the console.

105

1

106

1

h3. Bug Fixes

107

1

108

1

  * about 200 bugs have been fixed. For the list of closed and still open bugs, please look at

109

1

    https://code.ros.org/trac/opencv/report and

110

1

    http://sourceforge.net/tracker/?group_id=22870&atid=376677.

111

1

112

1

h3. Known issues

113

1

114

1

  * there are some sporadic test failures on different platforms.

115

1

    Most probably they are caused by some very special test cases

116

1

    (that are usually generated randomly on each test run), and the test cases

117

1

    are not properly handled by the functions or by the tests.

118

1

    Some of the tests have been reproduced and reported here:

119

1

120

1

    https://code.ros.org/trac/opencv/ticket/29

121

1

    https://code.ros.org/trac/opencv/ticket/113

122

1

    https://code.ros.org/trac/opencv/ticket/114

123

1

124

1

  * the new Python bindings do not include interface for the new C++ functionality and MLL.

125

1

    this is going to be addressed in some special intermediate OpenCV release

126

1

127

1

  * documentation is also incomplete at the moment and there are occasional formatting,

128

1

    grammar and semantical errors.

129

1

    We continue to improve it on a regular basis. Please, check the up-to-date online

130

1

    documentation at:

131

1

132

1

    http://opencv.willowgarage.com/documentation/c/index.html (C)

133

1

    http://opencv.willowgarage.com/documentation/cpp/index.html (C++)

134

1

    http://opencv.willowgarage.com/documentation/python/index.html (Python)

135

1

136

1

  * please also check the list of open bugs at

137

1

    https://code.ros.org/trac/opencv/report and

138

1

    http://sourceforge.net/tracker/?group_id=22870&atid=376677.

139

1

140

1

141

1

h2. 2.0

142

1

143

1

_*(September, 2009)*_

144

1

145

1

h3. New functionality

146

1

147

1

  # General:

148

1

149

1

    * Installation package on Windows has been fixed, MinGW is used for precompiled binaries (see the release notes on the Source``Forge site)

150

1

    * autotools-based build has been repaired

151

1

    * New Python interface officially in.

152

1

  # MLL:

153

1

154

1

    * The new-style class aliases (e.g. cv::SVM ~ CvSVM) and the train/predict methods, taking cv::Mat in addition to Cv``Mat, have been added. So now MLL can be used more seamlesly with the rest of the restyled OpenCV.

155

1

156

1

h3. Bug fixes

157

1

158

1

Thanks to everybody who submitted bug reports and/or patches!

159

1

160

1

   * A _lot_ of bugs present in 2.0 beta have been fixed, see the OpenCV bug tracker

161

1

   * http://sourceforge.net/tracker/?group_id=22870&atid=376677

162

163

164

165

h2. 2.0 beta

166

1

167

1

_*(September, 2009)*_

168

1

169

1

h3. New functionality

170

1

171

1

  * *General:*

172

1

173

1

    * The brand-new C++ interface for most of OpenCV functionality (cxcore, cv, highgui) has been introduced. Generally it means that you will need to do less coding to achieve the same results; it brings automatic memory management and many other advantages.

174

1

    ** See the C++ Reference section in opencv/doc/opencv.pdf and opencv/include/opencv/*.hpp.

175

1

    ** The previous interface is retained and still supported.

176

1

    * The source directory structure has been reorganized; now all the external headers are placed in the single directory on all platforms.

177

1

    * The primary build system is CMake,

178

1

    ** http://www.cmake.org (2.6.x is the preferable version).

179

1

    *** In Windows package the project files for Visual Studio, makefiles for MSVC,  Borland C++ or MinGW are not supplied anymore; please, generate them using CMake.

180

1

    *** In MacOSX the users can generate project files for Xcode.

181

1

    *** In Linux and any other platform the users can generate project files for  cross-platform IDEs, such as Eclipse or Code Blocks, or makefiles for building OpenCV from a command line.

182

1

    * OpenCV repository has been converted to Subversion, hosted at Source``Forge:

183

1

    *** http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary

184

1

    ** where the very latest snapshot is at

185

1

    *** http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/trunk,

186

1

    ** and the more or less stable version can be found at

187

1

    *** http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/tags/latest_tested_snapshot

188

1

189

1

  * *CXCORE, CV, CVAUX:*

190

1

191

1

    * CXCORE now uses Lapack (CLapack 3.1.1.1 in OpenCV 2.0) in its various linear algebra functions (such as solve, invert, SVD, determinant, eigen etc.) and the corresponding old-style functions (cvSolve, cvInvert etc.

192

1

    * Lots of new feature and object detectors and descriptors have been added (there is no documentation on them yet), see cv.hpp and cvaux.hpp:

193

1

    ** FAST - the fast corner detector, submitted by _*Edward Rosten*_

194

1

    ** MSER - maximally stable extremal regions, submitted by _*Liu Liu*_

195

1

    ** LDetector - fast circle-based feature detector

196

1

    *** _*by V. Lepetit (a.k.a. YAPE)*_

197

1

    ** Fern-based point classifier and the planar object detector -

198

1

    *** _*based on the works by M. Ozuysal and V. Lepetit*_

199

1

    ** One-way descriptor - a powerful PCA-based feature descriptor,

200

1

    *** _*S. Hinterstoisser, O. Kutter, N. Navab, P. Fua, and V. Lepetit, "Real-Time Learning of Accurate Patch Rectification"_*. Contributed by Victor Eruhimov

201

1

    ** Spin Images 3D feature descriptor

202

1

    *** based on the _*A. Johnson PhD thesis_*; implemented by _*Anatoly Baksheev*_

203

1

    ** Self-similarity features - contributed by _*Rainer Leinhar*_

204

1

    ** HOG people and object detector - the reimplementation of _*Navneet Dalal*_ framework

205

1

    *** (http://pascal.inrialpes.fr/soft/olt/). Currently, only the detection part is ported, but it is fully compatible with the original training code.

206

1

    *** See cvaux.hpp and opencv/samples/c/peopledetect.cpp.

207

1

    ** LBP (Local Binary Pattern) features

208

1

    ** Extended variant of the Haar feature-based object detector - implemented by _*Maria Dimashova_*. It now supports Haar features and LBPs, other features can be added in the same way.

209

1

    ** Adaptive skin detector and the fuzzy meanshift tracker - contributed by _*Farhad Dadgostar_*, see cvaux.hpp and opencv/samples/c/adaptiveskindetector.cpp

210

1

    * The new traincascade application complementing the new-style HAAR+LBP object detector has been added. See opencv/apps/traincascade.

211

1

    * The powerful library for approximate nearest neighbor search FLANN by _*Marius Muja*_ is now shipped with OpenCV, and the OpenCV-style interface to the library is included into cxcore. See cxcore.hpp and opencv/samples/c/find_obj.cpp

212

1

    * The bundle adjustment engine has been contributed by _*!PhaseSpace_*; see cvaux.hp

213

1

    * Added dense optical flow estimation function based on the paper

214

1

    ** _*"Two-Frame Motion Estimation Based on Polynomial Expansion" by G. Farnerback_*.

215

1

    ** See cv::calcOpticalFlowFarneback and the C++ documentation

216

1

    * Image warping operations (resize, remap, warpAffine, warpPerspective) now all support bicubic and Lanczos interpolation.

217

1

    * Most of the new linear and non-linear filtering operations (filter2D, sepFilter2D, erode, dilate ...) support arbitrary border modes and can use the valid image pixels outside of the ROI (i.e. the ROIs are not "isolated" anymore), see the C++ documentation.

218

1

    * The data can now be saved to and loaded from GZIP-compressed XML/YML files, e.g.: cvSave("a.xml.gz", my_huge_matrix);

219

1

220

1

  * *MLL:*

221

1

222

1

    * Added the Extremely Random Trees that train super-fast, comparing to Boosting or Random Trees (by _*Maria Dimashova_*).

223

1

    * The decision tree engine and based on it classes (Decision Tree itself, Boost, Random Trees) have been reworked and now:

224

1

    ** they consume much less memory (up to 200% savings)

225

1

    ** the training can be run in multiple threads (when OpenCV is built with OpenMP support)

226

1

    ** the boosting classification on numerical variables is especially fast because of the specialized low-overhead branch.

227

1

    * mltest has been added. While far from being complete, it contains correctness tests for some of the MLL classes.

228

1

229

1

  * *HighGUI:*

230

1

231

1

    * [Linux] The support for stereo cameras (currently Videre only) has been added.

232

1

    * There is now uniform interface for capturing video from two-, three- ... n-head cameras.

233

1

    * Images can now be compressed to or decompressed from buffers in the memory, see the C++ HighGUI reference manual

234

1

235

1

  * *Documentation:*

236

1

237

1

    * The reference manual has been converted from HTML to LaTeX (by _*James Bowman and Caroline Pantofaru_*), so there is now:

238

1

    ** opencv.pdf for reading offline

239

1

    ** and the online up-to-date documentation (as the result of LaTeX->Sphinx->HTML conversion) available at

240

1

    *** http://opencv.willowgarage.com/documentation/index.html

241

1

242

1

  * *Samples, misc.:*

243

1

244

1

    * Better eye detector has been contributed by _*Shiqi Yu_*, see opencv/data/haarcascades/*[lefteye|righteye]*.xml

245

1

    * sample LBP (Local Binary Pattern) cascade for the frontal face detection has been created by _*Maria Dimashova_*, see opencv/data/lbpcascades/lbpcascade_frontalface.xml

246

1

    * Several high-quality body parts and facial feature detectors have been

247

1

   ** _*contributed by Modesto Castrillon-Santana_*,

248

1

   ** see opencv/data/haarcascades/haarcascade_mcs*.xml

249

1

250

1

h3. Optimization

251

1

252

1

    * Many of the basic functions and the image processing operations(like arithmetic operations, geometric image transformations, filtering etc.) have got SSE2 optimization, so they are several times faster.

253

1

    * The model of IPP support has been changed. Now IPP is supposed to be detected by CMake at the configuration stage and linked against OpenCV. (In the beta it is not implemented yet though).

254

1

    * PNG encoder performance improved by factor of 4 by tuning the parameters

255

1

256

1

h3. Bug fixes

257

1

258

1

Many thanks to everybody who submitted bug reports and/or provided the patches!

259

1

260

1

 * see http://sourceforge.net/tracker/?group_id=22870&atid=376677 of the list of the closed and still opened bugs.

261

1

262

1

263

1

h3. Known issues

264

1

265

1

* configure+autotools based build is currently broken. Please, use CMake.

266

1

* OpenCV bug tracker at SF still lists about 150 open bugs. Some of them may be actually fixed already, and most of the remaining bugs are going to be fixed by OpenCV 2.0 gold.

267

1

* IPP is not supported. As the new OpenCV includes a lot of SSE2 code, it may be not such a serious problem, though. The support (at least for most important functions that do not have SSE2 optimization) will be returned in 2.0 gold.

268

1

* The documentation has been updated and improved a lot, but it still needs quite a bit of work:

269

1

** some of the new functionality in cvaux is not described yet.

270

1

** the bibliography part is broken

271

1

** there are quite a few known bugs and typos there

272

1

** many of the hyperlinks are not working.

273

1

* The existing tests partly cover the new functionality (via the old backward-compatibility OpenCV 1.x API), but the coverage is not sufficient of course.

274

1

* The new-style Python interface is not included yet

275

1

276

1

* Many of the problems will be addressed in 2.0 gold. If you have found some specific problem, please, put the record to the bug tracker:

277

1

278

1

   * http://sourceforge.net/tracker/?group_id=22870

279

1

   * Better if the bug reports will include a small code sample in C++/python + all the necessary data files needed to reproduce the problem.

280

1

281

1

282

1

h2. 1.1pre1

283

1

284

1

_*(October, 2008)*_

285

1

286

1

h3. New functionality

287

1

288

1

  * *General:*

289

1

290

1

    * Octave bindings have been added. See interfaces/swig/octave (for now, Linux only)

291

1

    * [Windows] OpenCV is now built with VS2005 with SSE2 and OpenMP support included (if you want to rebuild OpenCV using Express or Standard Edition of VS, use _make\opencv.vs2005.no_openmp.sln).

292

1

    * [Windows] Python bindings have been updated to use Python 2.6

293

1

    * [Windows] cvcam has been removed (as videoInput is now supported by highgui)

294

1

295

1

  * *CXCORE, CV, CVAUX:*

296

1

297

1

    * Speeded-up Robust Features (SURF), contributed by Liu Liu. see samples/c/find_obj.cpp and the documentation opencvref_cv.htm

298

1

    * Many improvements in camera calibration:

299

1

    ** Added stereo camera calibration: cvStereoCalibrate, cvStereoRectify etc.

300

1

    ** Single camera calibration now uses Levenberg-Marquardt method and supports extra flags to switch on/off optimization of individual camera parameters

301

1

    ** The optional 3rd radial distortion parameter (k3*r^6) is now supported in every calibration-related function

302

1

    * 2 stereo correspondence algorithms:

303

1

    ** very fast block matching method by Kurt Konolige (processes the Tsukuba stereo pair in <10ms on Core2``Duo laptop)

304

1

    ** slow but more accurate graph-cut based algorithm by Kolmogorov and Zabin

305

1

    * Better homography estimation algorithms (RANSAC and LMEDs)

306

1

    * new C++ template image classes contributed by Daniel Filip (Google inc.). see opencv/cxcore/include/cvwimage.h

307

1

    * Fast approximate nearest neighbor search (by Xavier Delacour)

308

1

    * Codebook method for background/foreground segmentation (by Gary Bradski)

309

1

    * Sort function (contributed by Shiqi Yu)

310

1

    * [OpenCV+IPP] Face Detection (cvHaarDetectObjects) now runs much faster (up to 2x faster) when using IPP 5.3 or higher.

311

1

    * Much faster (~4x faster) fixed-point variant of cvRemap has been added

312

1

313

1

  * *MLL:*

314

1

315

1

    * Python bindings for MLL have been added. There are no samples yet.

316

1

317

1

  * *HighGUI:*

318

1

319

1

    * [Windows, 32bit] Added support for videoInput library. Hence, cvcam is [almost] not needed anymore

320

1

    * [Windows, 32bit] FFMPEG can now be used for video decoding/encoding via ffopencv*.dll

321

1

    * [Linux] Added unicap support

322

1

    * Improved internal video capturing and video encoding APIs

323

1

324

1

  * *Documentation:*

325

1

326

1

    * OpenCV book has been published (sold separately :) see docs/index.htm)

327

1

328

1

  * *New samples (opencv/samples):*

329

1

330

1

    * Many Octave samples

331

1

    * find_obj.cpp (SURF), bgfg_codebook.cpp (Codebook BG/FG segmentation),

332

1

      stereo_calib.cpp (Stereo calibration and stereo correspondence)

333

1

334

1

h3. Bug fixes

335

1

336

1

Many thanks to everybody who submitted bug reports and/or provided the patches!

337

1

338

1

    * added dma_unlisten to icvCloseCAM_DC1394 (thanks to Victor Benso)

339

1

    * LMEDs algorithm for cvFindFundamentalMat estimation has been fixed

340

1

    * Broken debug build of highgui in VS2005/2008 (SF #2056185, SF #1709435)

341

1

    * cvFindChessboardCorners memory leak and incorrect size checks (SF #1972896, SF #1910323)

342

1

    * disabling GTK causes v4l runtime error (SF #2088102)

343

1

    * cvsetmousecallback bug (SF #2053529)

344

1

    * libhighgui needed deprecated "img_convert" replacement (SF #2011753)

345

1

    * Segfault in createsamples caused by uninitialized variable (SF #1977989)

346

1

    * Data Alignment Issue in bgfg_gaussmix (SF #1961755)

347

1

    * libpng need to be updated (SF #1952793)

348

1

    * cvCreateVideoWriter_Win32 - identifier not found (SF #1944254)

349

1

    * Bug in cvmorph.cpp (SF #1908844)

350

1

    * dilate (cvDilate) works bogus with default kernel (SF #1887130)

351

1

    * CvEM non-default constructor declared but not defined (SF #1830346)

352

1

    * cvFloodFill (in ver 1.0) Hangs (SF #1790565)

353

1

    * double delete in Cv``Image (SF #1733437)

354

1

    * cvFilter2D anchor default value is not working properly (SF #1713779)

355

1

    * cvCvtColor - Bug? in converting HSV2RGB (SF #1676344)

356

1

    * Invalid selection of the MKL-dll version in cvUseOptimized() (SF #1633017)

357

1

    * ...

358

1

359

1

h3. Known issues

360

1

361

1

    * Borland compiler is not supported (but might work) in this update.

362

1

    * 64-bit Windows is not supported (but might work) in this update.

363

1

    * SF bug tracker still contains over 50 records of open bugs.

364

1

      Many of them will be addressed in the next update.

365

1

366

1

367

1

h2. 1.0

368

1

369

1

_*(October, 2006)*_

370

1

371

1

h3. New functionality

372

1

373

1

  * *General:*

374

1

375

1

    * Windows installation package now includes precompiled Python module

376

1

    * Borland C++ (v5.6+) makefiles re-added (see _make\make_all_bc.mak)

377

1

378

1

  * *CV:*

379

1

380

1

    * Inpainting was added, see samples/c/inpaint.cpp (thanks to Alexander Kibkalo and Alexey Kalinichenko)

381

1

382

1

  * *MLL:*

383

1

384

1

    * Boosted tree classifiers added, see MLL reference manual

385

1

386

1

  * *HighGUI:*

387

1

388

1

    * Jpeg2000 and EXR support in highgui using JasPer and OpenEXR (thanks to Nils Hasler) OpenEXR is not built-in to highgui on Windows by default. See otherlibs/_graphics/readme.txt.

389

1

    * >8-bit image input/output for PNG, Jpeg2000 and OpenEXR (thanks to Nils Hasler)

390

1

    * Updated CMUcamera wrapper.

391

1

    * Several V4L2 improvements and fixes

392

1

393

1

  * *New samples (opencv/samples):*

394

1

395

1

    * Camera calibration sample: calibrate.cpp

396

1

    * Inpainting sample: inpaint.cpp

397

1

    * Letter recognition using the existing database with random trees/boosted trees/neural nets: leter_recog.cpp

398

1

    * A few new Python samples.

399

1

400

1

  * *Tests:*

401

1

402

1

    * Most old cv tests have been rewritten/converted to use cxts instead of trs. TRS is not used by the tests anymore.

403

1

404

1

h3. Bug fixes

405

1

406

1

Many thanks to everybody who submitted bug reports and/or provided the patches!

407

1

408

1

     * Incorrect normalization in cvFindFundamentalMat (SF #1550855)

409

1

     * Occasional cvGetQuadrangleSubPix crashes (SF #1549168)

410

1

     * Occasional cvCalcOpticalFlowPyrLK crashes

411

1

     * Internal GCC 3.x error in cvFloor/cvCeil (SF #1546657)

412

1

     * Several Python-related bugs (SF #1570109, SF #1572800, SF #1543593)

413

1

     * Bugs in contour processing/rendering (SF #1550889,

414

1

     * x86-64 related bugs (SF #1526133, SF #1527997)

415

1

     * OpenCV dispatcher tried to load old versions of IPP (pre 4.x) that caused many failures (see SF #1555093, for example)

416

1

     * Inaccurate results in some cases in cvFitEllipse, cvFitLine, cvHoughCircles.

417

1

     * Severe bug in Bayesian classifier, broken regression tree in MLL

418

1

     * Overflow in blurring filter

419

1

     * cvSaveImage() modified saved images leading to potential problems in threaded apps.

420

1

     * Numeric values in XML/YAML were emitted/parsed incorrectly on some non-US locales in Linux

421

1

     * VS2005 Express/Standard Editions could not build OpenCV (due to missing OpenMP support)

422

1

423

1

h3. Known issues

424

1

425

1

    * Certain tests may fail on IA64 (Itanium), e.g.

426

1

      gaussian filter, fundamental matrix estimation, chessboard detector.

Login	Password

ChangeLog v10-v21