3 <title>OpenCV ChangeLog</title>
4 <style TYPE="text/css">
16 <p class="Blurb"><i>(April, 2010)</i></p>
19 OpenCV 2.1 is basically a stabilized OpenCV 2.0, yet there are a few new features.
21 >>> General modifications
23 - SVN repository has been migrated from SourceForge to https://code.ros.org/svn/opencv.
24 The bug tracker has been moved to https://code.ros.org/trac/opencv/.
25 And we also have OpenCV twitter! http://twitter.com/opencvlibrary
27 - The whole OpenCV is now using exceptions instead of the old libc-style mechanism.
28 That is, instead of checking error code with cvGetErrStatus() (which currently always returns 0)
29 you can now just call OpenCV functions inside C++ try-catch statements,
30 cv::Exception is now derived from std::exception.
32 - OpenCV does not support autotools-based build scripts,
33 CMake (www.cmake.org) is the only way to build OpenCV on any OS.
34 See http://opencv.willowgarage.com/wiki/InstallGuide.
36 - All the parallel loops in OpenCV have been converted from OpenMP
37 to Intel TBB (http://www.threadingbuildingblocks.org/). Thus parallel version of OpenCV
38 can now be built using MSVC 2008 Express Edition or using earlier than 4.2 versions of GCC.
40 - SWIG-based Python wrappers are still included,
41 but they are not built by default and it's generally preferable to use the new wrappers.
42 The python samples have been rewritten by James Bowman to use the new-style Python wrappers,
43 which have been also created by James.
45 - OpenCV can now be built and run in 64-bit mode on MacOSX 10.6 and Windows (see HighGUI and known problems below).
46 On Windows both MSVC 2008 and mingw64 are known to work.
48 - In theory OpenCV is now able to determine the host CPU on-fly and make use of SSE/SSE2/... instructions,
49 if they are available. That is, it should be more safe to use WITH_SSE* flags in CMake.
50 However, if you want maximum portability, it's recommended to turn on just WITH_SSE and WITH_SSE2
51 and leave other SSE* turned off, as we found that using WITH_SSE3, WITH_SSSE3 and WITH_SSE4_1 can yield
52 the code incompatible with Intel's pre-Penryn or AMD chips.
54 >>> New functionality, features:
58 * Grabcut (http://en.wikipedia.org/wiki/GrabCut) image segmentation algorithm has been implemented.
59 See opencv/samples/c/grabcut.cpp
61 * new improved version of one-way descriptor is added. See opencv/samples/c/one_way_sample.cpp
63 * modified version of H. Hirschmuller semi-global stereo matching algorithm that we call SGBM
64 (semi-global block matching) has been created. It is much faster than Kolmogorov's graph
65 cuts-based algorithm and yet it's usually better than the block matching StereoBM algorithm.
66 See opencv/samples/c/stereo_matching.cpp.
68 * existing StereoBM stereo correspondence algorithm by K. Konolige was noticeably improved:
69 added the optional left-right consistency check and speckle filtering,
70 improved performance (by ~20%).
72 * User can now control the image areas visible after the stereo rectification
73 (see the extended stereoRectify/cvStereoRectify), and also limit the region
74 where the disparity is computed (see CvStereoBMState::roi1, roi2; getValidDisparityROI).
76 * Mixture-of-Gaussian based background subtraction algorithm has been rewritten for better performance
77 and better accuracy. Alternative C++ interface BackgroundSubtractor has been provided,
78 along with the possibility to use the trained background model to segment the foreground
79 without updating the model. See opencv/samples/c/bgfg_segm.cpp.
83 * MacOSX: OpenCV now includes Cocoa and QTKit backends, in addition to Carbon and Quicktime.
84 Therefore you can build OpenCV as 64-bit library. Thanks to Andre Cohen and Nicolas Butko, which components
86 Note however that the backend are now in the alpha state, they can crash or leak memory,
87 so for anything more serious than quick experiments you may prefer to use Carbon and Quicktime.
88 To do that, pass USE_CARBON=ON and USE_QUICKTIME=ON to CMake and build OpenCV in 32-bit mode
89 (i.e. select i386 architecture in Xcode).
91 * Windows. OpenCV can now be built in 64-bit mode with MSVC 2008 and also mingw64.
93 * Fullscreen has been added (thanks to Yannick Verdie).
94 Call cvSetWindowProperty(window_name, CV_WINDOW_FULLSCREEN, 1) to make the particular window
95 to fill the whole screen. This feature is not supported in the Cocoa bindings yet.
97 * gstreamer backend has been improved a lot (thanks to Stefano Fabri)
101 - A few dozens of new tests have been written and many existing tests have been extended
102 to verify OpenCV correctness thoroughly. As a result, we brought the test coverage from
103 rather mediocre numbers to pretty impressive ones (especially for cxcore and cv)!
105 Module OpenCV 2.0 coverage OpenCV 2.1
106 (functions/conditions) (functions/conditions)
113 - Many new regression tests have been written in Python that check both OpenCV and the new-style bindings.
115 - The test data moved to the separate repository: https://code.ros.org/svn/opencv/trunk/opencv_extra/testdata.
116 And it is not included into the package, thus some tests from the cvtest and mltest will report about the missing data.
117 You can download the directory to your hard drive and run cvtest like:
118 ./cvtest -d <path_to_opencv_extra>/testdata/cv
119 ./mltest -d <path_to_opencv_extra>/testdata/ml
121 - The test engine has been improved:
122 added flags -tn, -seed, -r
123 the detailed information about failed tests is displayed right in the console.
127 - about 200 bugs have been fixed. For the list of closed and still open bugs, please look at
128 https://code.ros.org/trac/opencv/report and
129 http://sourceforge.net/tracker/?group_id=22870&atid=376677.
131 >>> Known problems/limitations.
133 - there are some sporadic test failures on different platforms.
134 Most probably they are caused by some very special test cases
135 (that are usually generated randomly on each test run) and the test cases
136 are not properly handled by the functions or by the tests.
137 Some of the tests have been reproduced and reported here:
139 https://code.ros.org/trac/opencv/ticket/29
140 https://code.ros.org/trac/opencv/ticket/113
141 https://code.ros.org/trac/opencv/ticket/114
143 - the new Python bindings do not include interface for the new C++ functionality and MLL.
144 this is going to be addressed in some special intermediate OpenCV release
146 - documentation is also incomplete at the moment and there are occasional formatting,
147 grammar and semantical errors.
148 We continue to improve it on a regular basis. Please, check the up-to-date online
151 http://opencv.willowgarage.com/documentation/c/index.html (C)
152 http://opencv.willowgarage.com/documentation/cpp/index.html (C++)
153 http://opencv.willowgarage.com/documentation/python/index.html (Python)
155 - please also check the list of open bugs at
156 https://code.ros.org/trac/opencv/report and
157 http://sourceforge.net/tracker/?group_id=22870&atid=376677.
161 <hr><h2>2.0 beta</h2>
162 <p class="Blurb"><i>(September, 2009)</i></p>
165 >>> New functionality, features: <<<
168 * The brand-new C++ interface for most of OpenCV functionality
169 (cxcore, cv, highgui) has been introduced.
170 Generally it means that you will need to do less coding to achieve the same results;
171 it brings automatic memory management and many other advantages.
172 See the C++ Reference section in opencv/doc/opencv.pdf and opencv/include/opencv/*.hpp.
173 The previous interface is retained and still supported.
175 * The source directory structure has been reogranized; now all the external headers are placed
176 in the single directory on all platforms.
178 * The primary build system is CMake, http://www.cmake.org (2.6.x is the preferable version).
179 + In Windows package the project files for Visual Studio, makefiles for MSVC,
180 Borland C++ or MinGW are note supplied anymore; please, generate them using CMake.
182 + In MacOSX the users can generate project files for Xcode.
184 + In Linux and any other platform the users can generate project files for
185 cross-platform IDEs, such as Eclipse or Code Blocks,
186 or makefiles for building OpenCV from a command line.
188 * OpenCV repository has been converted to Subversion, hosted at SourceForge:
189 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary
190 where the very latest snapshot is at
191 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/trunk,
192 and the more or less stable version can be found at
193 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/tags/latest_tested_snapshot
197 * CXCORE now uses Lapack (CLapack 3.1.1.1 in OpenCV 2.0) in its various linear algebra functions
198 (such as solve, invert, SVD, determinant, eigen etc.) and the corresponding old-style functions
199 (cvSolve, cvInvert etc.)
201 * Lots of new feature and object detectors and descriptors have been added
202 (there is no documentation on them yet), see cv.hpp and cvaux.hpp:
204 + FAST - the fast corner detector, submitted by Edward Rosten
206 + MSER - maximally stable extremal regions, submitted by Liu Liu
208 + LDetector - fast circle-based feature detector by V. Lepetit (a.k.a. YAPE)
210 + Fern-based point classifier and the planar object detector -
211 based on the works by M. Ozuysal and V. Lepetit
213 + One-way descriptor - a powerful PCA-based feature descriptor,
214 (S. Hinterstoisser, O. Kutter, N. Navab, P. Fua, and V. Lepetit,
215 "Real-Time Learning of Accurate Patch Rectification").
216 Contributed by Victor Eruhimov
218 + Spin Images 3D feature descriptor - based on the A. Johnson PhD thesis;
219 implemented by Anatoly Baksheev
221 + Self-similarity features - contributed by Rainer Leinhart
223 + HOG people and object detector - the reimplementation of Navneet Dalal framework
224 (http://pascal.inrialpes.fr/soft/olt/). Currently, only the detection part is ported,
225 but it is fully compatible with the original training code.
226 See cvaux.hpp and opencv/samples/c/peopledetect.cpp.
228 + Extended variant of the Haar feature-based object detector - implemented by Maria Dimashova.
229 It now supports Haar features and LBPs (local binary patterns);
230 other features can be more or less easily added
232 + Adaptive skin detector and the fuzzy meanshift tracker - contributed by Farhad Dadgostar,
233 see cvaux.hpp and opencv/samples/c/adaptiveskindetector.cpp
235 * The new traincascade application complementing the new-style HAAR+LBP object detector has been added.
236 See opencv/apps/traincascade.
238 * The powerful library for approximate nearest neighbor search FLANN by Marius Muja
239 is now shipped with OpenCV, and the OpenCV-style interface to the library
240 is included into cxcore. See cxcore.hpp and opencv/samples/c/find_obj.cpp
242 * The bundle adjustment engine has been contributed by PhaseSpace; see cvaux.hpp
244 * Added dense optical flow estimation function (based on the paper
245 "Two-Frame Motion Estimation Based on Polynomial Expansion" by G. Farnerback).
246 See cv::calcOpticalFlowFarneback and the C++ documentation
248 * Image warping operations (resize, remap, warpAffine, warpPerspective)
249 now all support bicubic and Lanczos interpolation.
251 * Most of the new linear and non-linear filtering operations (filter2D, sepFilter2D, erode, dilate ...)
252 support arbitrary border modes and can use the valid image pixels outside of the ROI
253 (i.e. the ROIs are not "isolated" anymore), see the C++ documentation.
255 * The data can now be saved to and loaded from GZIP-compressed XML/YML files, e.g.:
256 cvSave("a.xml.gz", my_huge_matrix);
259 * Added the Extremely Random Trees that train super-fast,
260 comparing to Boosting or Random Trees (by Maria Dimashova).
262 * The decision tree engine and based on it classes
263 (Decision Tree itself, Boost, Random Trees)
264 have been reworked and now:
265 + they consume much less memory (up to 200% savings)
266 + the training can be run in multiple threads (when OpenCV is built with OpenMP support)
267 + the boosting classification on numerical variables is especially
268 fast because of the specialized low-overhead branch.
270 * mltest has been added. While far from being complete,
271 it contains correctness tests for some of the MLL classes.
274 * [Linux] The support for stereo cameras (currently Videre only) has been added.
275 There is now uniform interface for capturing video from two-, three- ... n-head cameras.
277 * Images can now be compressed to or decompressed from buffers in the memory,
278 see the C++ HighGUI reference manual
281 * The reference manual has been converted from HTML to LaTeX (by James Bowman and Caroline Pantofaru),
283 + opencv.pdf for reading offline
284 + and the online up-to-date documentation
285 (as the result of LaTeX->Sphinx->HTML conversion) available at
286 http://opencv.willowgarage.com/documentation/index.html
289 * Better eye detector has been contributed by Shiqi Yu,
290 see opencv/data/haarcascades/*[lefteye|righteye]*.xml
291 * sample LBP cascade for the frontal face detection
292 has been created by Maria Dimashova,
293 see opencv/data/lbpcascades/lbpcascade_frontalface.xml
294 * Several high-quality body parts and facial feature detectors
295 have been contributed by Modesto Castrillon-Santana,
296 see opencv/data/haarcascades/haarcascade_mcs*.xml
299 * Many of the basic functions and the image processing operations
300 (like arithmetic operations, geometric image transformations, filtering etc.)
301 have got SSE2 optimization, so they are several times faster.
303 - The model of IPP support has been changed. Now IPP is supposed to be
304 detected by CMake at the configuration stage and linked against OpenCV.
305 (In the beta it is not implemented yet though).
307 * PNG encoder performance improved by factor of 4 by tuning the parameters
311 (see http://sourceforge.net/tracker/?group_id=22870&atid=376677 of the list
312 of the closed and still opened bugs).
314 Many thanks to everybody who submitted bug reports and/or provided the patches!
317 * configure+autotools based build is currently broken.
319 * OpenCV bug tracker at SF still lists about 150 open bugs.
320 Some of them may be actually fixed already, and most of the remaining bugs
321 are going to be fixed by OpenCV 2.0 gold.
322 * IPP is not supported. As the new OpenCV includes a lot of SSE2 code,
323 it may be not such a serious problem, though.
324 The support (at least for most important functions that do not have
325 SSE2 optimization) will be returned in 2.0 gold.
326 * The documentation has been updated and improved a lot, but it still
327 needs quite a bit of work:
328 - some of the new functionality in cvaux is not described yet.
329 - the bibliography part is broken
330 - there are quite a few known bugs and typos there
331 - many of the hyperlinks are not working.
332 * The existing tests partly cover the new functionality
333 (via the old backward-compatibility OpenCV 1.x API), but the coverage is
334 not sufficient of course.
335 * The new-style Python interface is not included yet
337 Many of the problems will be addressed in 2.0 gold.
338 If you have found some specific problem, please, put the record to the bug tracker:
339 http://sourceforge.net/tracker/?group_id=22870
340 Better if the bug reports will include a small code sample in C++/python +
341 all the necessary data files needed to reproduce the problem.
346 <p class="Blurb"><i>(September, 2009)</i></p>
349 >>> New functionality, features:
352 * Installation package on Windows has been fixed,
353 MinGW is used for precompiled binaries
354 (see the release notes on the SourceForge site)
356 * autotools-based build has been repaired
359 * The new-style class aliases (e.g. cv::SVM ~ CvSVM) and the train/predict methods,
360 taking cv::Mat in addition to CvMat, have been added. So now MLL can be used
361 more seamlesly with the rest of the restyled OpenCV.
365 - A _lot_ of bugs present in 2.0 beta have been fixed, see the OpenCV bug tracker
366 http://sourceforge.net/tracker/?group_id=22870&atid=376677
367 Thanks to everybody who submitted bug reports and/or patches!
370 <hr><h2>2.0 beta</h2>
371 <p class="Blurb"><i>(September, 2009)</i></p>
374 >>> New functionality, features: <<<
377 * The brand-new C++ interface for most of OpenCV functionality
378 (cxcore, cv, highgui) has been introduced.
379 Generally it means that you will need to do less coding to achieve the same results;
380 it brings automatic memory management and many other advantages.
381 See the C++ Reference section in opencv/doc/opencv.pdf and opencv/include/opencv/*.hpp.
382 The previous interface is retained and still supported.
384 * The source directory structure has been reogranized; now all the external headers are placed
385 in the single directory on all platforms.
387 * The primary build system is CMake, http://www.cmake.org (2.6.x is the preferable version).
388 + In Windows package the project files for Visual Studio, makefiles for MSVC,
389 Borland C++ or MinGW are note supplied anymore; please, generate them using CMake.
391 + In MacOSX the users can generate project files for Xcode.
393 + In Linux and any other platform the users can generate project files for
394 cross-platform IDEs, such as Eclipse or Code Blocks,
395 or makefiles for building OpenCV from a command line.
397 * OpenCV repository has been converted to Subversion, hosted at SourceForge:
398 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary
399 where the very latest snapshot is at
400 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/trunk,
401 and the more or less stable version can be found at
402 http://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/tags/latest_tested_snapshot
406 * CXCORE now uses Lapack (CLapack 3.1.1.1 in OpenCV 2.0) in its various linear algebra functions
407 (such as solve, invert, SVD, determinant, eigen etc.) and the corresponding old-style functions
408 (cvSolve, cvInvert etc.)
410 * Lots of new feature and object detectors and descriptors have been added
411 (there is no documentation on them yet), see cv.hpp and cvaux.hpp:
413 + FAST - the fast corner detector, submitted by Edward Rosten
415 + MSER - maximally stable extremal regions, submitted by Liu Liu
417 + LDetector - fast circle-based feature detector by V. Lepetit (a.k.a. YAPE)
419 + Fern-based point classifier and the planar object detector -
420 based on the works by M. Ozuysal and V. Lepetit
422 + One-way descriptor - a powerful PCA-based feature descriptor,
423 (S. Hinterstoisser, O. Kutter, N. Navab, P. Fua, and V. Lepetit,
424 "Real-Time Learning of Accurate Patch Rectification").
425 Contributed by Victor Eruhimov
427 + Spin Images 3D feature descriptor - based on the A. Johnson PhD thesis;
428 implemented by Anatoly Baksheev
430 + Self-similarity features - contributed by Rainer Leinhart
432 + HOG people and object detector - the reimplementation of Navneet Dalal framework
433 (http://pascal.inrialpes.fr/soft/olt/). Currently, only the detection part is ported,
434 but it is fully compatible with the original training code.
435 See cvaux.hpp and opencv/samples/c/peopledetect.cpp.
437 + Extended variant of the Haar feature-based object detector - implemented by Maria Dimashova.
438 It now supports Haar features and LBPs (local binary patterns);
439 other features can be more or less easily added
441 + Adaptive skin detector and the fuzzy meanshift tracker - contributed by Farhad Dadgostar,
442 see cvaux.hpp and opencv/samples/c/adaptiveskindetector.cpp
444 * The new traincascade application complementing the new-style HAAR+LBP object detector has been added.
445 See opencv/apps/traincascade.
447 * The powerful library for approximate nearest neighbor search FLANN by Marius Muja
448 is now shipped with OpenCV, and the OpenCV-style interface to the library
449 is included into cxcore. See cxcore.hpp and opencv/samples/c/find_obj.cpp
451 * The bundle adjustment engine has been contributed by PhaseSpace; see cvaux.hpp
453 * Added dense optical flow estimation function (based on the paper
454 "Two-Frame Motion Estimation Based on Polynomial Expansion" by G. Farnerback).
455 See cv::calcOpticalFlowFarneback and the C++ documentation
457 * Image warping operations (resize, remap, warpAffine, warpPerspective)
458 now all support bicubic and Lanczos interpolation.
460 * Most of the new linear and non-linear filtering operations (filter2D, sepFilter2D, erode, dilate ...)
461 support arbitrary border modes and can use the valid image pixels outside of the ROI
462 (i.e. the ROIs are not "isolated" anymore), see the C++ documentation.
464 * The data can now be saved to and loaded from GZIP-compressed XML/YML files, e.g.:
465 cvSave("a.xml.gz", my_huge_matrix);
468 * Added the Extremely Random Trees that train super-fast,
469 comparing to Boosting or Random Trees (by Maria Dimashova).
471 * The decision tree engine and based on it classes
472 (Decision Tree itself, Boost, Random Trees)
473 have been reworked and now:
474 + they consume much less memory (up to 200% savings)
475 + the training can be run in multiple threads (when OpenCV is built with OpenMP support)
476 + the boosting classification on numerical variables is especially
477 fast because of the specialized low-overhead branch.
479 * mltest has been added. While far from being complete,
480 it contains correctness tests for some of the MLL classes.
483 * [Linux] The support for stereo cameras (currently Videre only) has been added.
484 There is now uniform interface for capturing video from two-, three- ... n-head cameras.
486 * Images can now be compressed to or decompressed from buffers in the memory,
487 see the C++ HighGUI reference manual
490 * The reference manual has been converted from HTML to LaTeX (by James Bowman and Caroline Pantofaru),
492 + opencv.pdf for reading offline
493 + and the online up-to-date documentation
494 (as the result of LaTeX->Sphinx->HTML conversion) available at
495 http://opencv.willowgarage.com/documentation/index.html
498 * Better eye detector has been contributed by Shiqi Yu,
499 see opencv/data/haarcascades/*[lefteye|righteye]*.xml
500 * sample LBP cascade for the frontal face detection
501 has been created by Maria Dimashova,
502 see opencv/data/lbpcascades/lbpcascade_frontalface.xml
503 * Several high-quality body parts and facial feature detectors
504 have been contributed by Modesto Castrillon-Santana,
505 see opencv/data/haarcascades/haarcascade_mcs*.xml
508 * Many of the basic functions and the image processing operations
509 (like arithmetic operations, geometric image transformations, filtering etc.)
510 have got SSE2 optimization, so they are several times faster.
512 - The model of IPP support has been changed. Now IPP is supposed to be
513 detected by CMake at the configuration stage and linked against OpenCV.
514 (In the beta it is not implemented yet though).
516 * PNG encoder performance improved by factor of 4 by tuning the parameters
520 (see http://sourceforge.net/tracker/?group_id=22870&atid=376677 of the list
521 of the closed and still opened bugs).
523 Many thanks to everybody who submitted bug reports and/or provided the patches!
526 * configure+autotools based build is currently broken.
528 * OpenCV bug tracker at SF still lists about 150 open bugs.
529 Some of them may be actually fixed already, and most of the remaining bugs
530 are going to be fixed by OpenCV 2.0 gold.
531 * IPP is not supported. As the new OpenCV includes a lot of SSE2 code,
532 it may be not such a serious problem, though.
533 The support (at least for most important functions that do not have
534 SSE2 optimization) will be returned in 2.0 gold.
535 * The documentation has been updated and improved a lot, but it still
536 needs quite a bit of work:
537 - some of the new functionality in cvaux is not described yet.
538 - the bibliography part is broken
539 - there are quite a few known bugs and typos there
540 - many of the hyperlinks are not working.
541 * The existing tests partly cover the new functionality
542 (via the old backward-compatibility OpenCV 1.x API), but the coverage is
543 not sufficient of course.
544 * The new-style Python interface is not included yet
546 Many of the problems will be addressed in 2.0 gold.
547 If you have found some specific problem, please, put the record to the bug tracker:
548 http://sourceforge.net/tracker/?group_id=22870
549 Better if the bug reports will include a small code sample in C++/python +
550 all the necessary data files needed to reproduce the problem.
554 <p class="Blurb"><i>(October, 2008)</i></p>
558 >>> New functionality/features: <<<
561 * Octave bindings have been added. See interfaces/swig/octave (for now, Linux only)
562 * [Windows] OpenCV is now built with VS2005 with SSE2 and OpenMP support
563 included (if you want to rebuild OpenCV using Express or
564 Standard Edition of VS, use _make\opencv.vs2005.no_openmp.sln).
565 * [Windows] Python bindings have been updated to use Python 2.6
566 * [Windows] cvcam has been removed (as videoInput is now supported by highgui)
569 * Speeded-up Robust Features (SURF), contributed by Liu Liu.
570 see samples/c/find_obj.cpp and the documentation opencvref_cv.htm
571 * Many improvements in camera calibration:
572 - Added stereo camera calibration: cvStereoCalibrate, cvStereoRectify etc.
573 - Single camera calibration now uses Levenberg-Marquardt method and supports
574 extra flags to switch on/off optimization of individual camera parameters
575 - The optional 3rd radial distortion parameter (k3*r^6) is now supported in
576 every calibration-related function
577 * 2 stereo correspondence algorithms:
578 - very fast block matching method by Kurt Konolige
579 (processes the Tsukuba stereo pair in <10ms on Core2Duo laptop)
580 - slow but more accurate graph-cut based algorithm by Kolmogorov and Zabin
581 * Better homography estimation algorithms (RANSAC and LMEDs)
582 * new C++ template image classes contributed by Daniel Filip (Google inc.).
583 see opencv/cxcore/include/cvwimage.h
584 * Fast approximate nearest neighbor search (by Xavier Delacour)
585 * Codebook method for background/foreground segmentation (by Gary Bradski)
586 * Sort function (contributed by Shiqi Yu)
587 * [OpenCV+IPP] Face Detection (cvHaarDetectObjects) now runs much faster
588 (up to 2x faster) when using IPP 5.3 or higher.
589 * Much faster (~4x faster) fixed-point variant of cvRemap has been added
592 * Python bindings for MLL have been added. There are no samples yet.
595 * [Windows, 32bit] Added support for videoInput library.
596 Hence, cvcam is [almost] not needed anymore
597 * [Windows, 32bit] FFMPEG can now be used for video decoding/encoding
599 * [Linux] Added unicap support
600 * Improved internal video capturing and video encoding APIs
603 * OpenCV book has been published (sold separately :) see docs/index.htm)
605 - New samples (opencv/samples):
606 * Many Octave samples
607 * find_obj.cpp (SURF), bgfg_codebook.cpp (Codebook BG/FG segmentation),
608 stereo_calib.cpp (Stereo calibration and stereo correspondence)
611 Many thanks to everybody who submitted bug reports and/or provided the patches!
613 * added dma_unlisten to icvCloseCAM_DC1394 (thanks to Victor Benso)
614 * LMEDs algorithm for cvFindFundamentalMat estimation has been fixed
615 * Broken debug build of highgui in VS2005/2008 (SF #2056185, SF #1709435)
616 * cvFindChessboardCorners memory leak and incorrect size checks
617 (SF #1972896, SF #1910323)
618 * disabling GTK causes v4l runtime error (SF #2088102)
619 * cvsetmousecallback bug (SF #2053529)
620 * libhighgui needed deprecated "img_convert" replacement (SF #2011753)
621 * Segfault in createsamples caused by uninitialized variable (SF #1977989)
622 * Data Alignment Issue in bgfg_gaussmix (SF #1961755)
623 * libpng need to be updated (SF #1952793)
624 * cvCreateVideoWriter_Win32 - identifier not found (SF #1944254)
625 * Bug in cvmorph.cpp (SF #1908844)
626 * dilate (cvDilate) works bogus with default kernel (SF #1887130)
627 * CvEM non-default constructor declared but not defined (SF #1830346)
628 * cvFloodFill (in ver 1.0) Hangs (SF #1790565)
629 * double delete in CvImage (SF #1733437)
630 * cvFilter2D anchor default value is not working properly (SF #1713779)
631 * cvCvtColor - Bug? in converting HSV2RGB (SF #1676344)
632 * Invalid selection of the MKL-dll version in cvUseOptimized() (SF #1633017)
636 * Borland compiler is not supported (but might work) in this update.
637 * 64-bit Windows is not supported (but might work) in this update.
638 * SF bug tracker still contains over 50 records of open bugs.
639 Many of them will be addressed in the next update.
643 <p class="Blurb"><i>(October, 2006)</i></p>
646 >>> New functionality/features: <<<
649 * Windows installation package now includes precompiled Python module
650 * Borland C++ (v5.6+) makefiles re-added (see _make\make_all_bc.mak)
653 * Inpainting was added, see samples/c/inpaint.cpp
654 (thanks to Alexander Kibkalo and Alexey Kalinichenko)
657 * Boosted tree classifiers added, see MLL reference manual
660 * Jpeg2000 and EXR support in highgui using JasPer and OpenEXR (thanks to Nils Hasler)
661 OpenEXR is not built-in to highgui on Windows by default. See
662 otherlibs/_graphics/readme.txt.
663 * >8-bit image input/output for PNG, Jpeg2000 and OpenEXR (thanks to Nils Hasler)
664 * Updated CMUcamera wrapper.
665 * Several V4L2 improvements and fixes
667 - New samples (opencv/samples):
668 * Camera calibration sample: calibrate.cpp
669 * Inpainting sample: inpaint.cpp
670 * Letter recognition using the existing database with
671 random trees/boosted trees/neural nets: leter_recog.cpp
672 * A few new Python samples.
675 * Most old cv tests have been rewritten/converted to use cxts instead of trs.
676 TRS is not used by the tests anymore.
679 Many thanks to everybody who submitted bug reports and/or provided the patches!
681 * Incorrect normalization in cvFindFundamentalMat (SF #1550855)
682 * Occasional cvGetQuadrangleSubPix crashes (SF #1549168)
683 * Occasional cvCalcOpticalFlowPyrLK crashes
684 * Internal GCC 3.x error in cvFloor/cvCeil (SF #1546657)
685 * Several Python-related bugs (SF #1570109, SF #1572800, SF #1543593)
686 * Bugs in contour processing/rendering (SF #1550889,
687 * x86-64 related bugs (SF #1526133, SF #1527997)
688 * OpenCV dispatcher tried to load old versions of IPP (pre 4.x)
689 that caused many failures (see SF #1555093, for example)
690 * Inaccurate results in some cases in cvFitEllipse, cvFitLine, cvHoughCircles.
691 * Severe bug in Bayesian classifier, broken regression tree in MLL
692 * Overflow in blurring filter
693 * cvSaveImage() modified saved images leading to potential problems in threaded apps.
694 * Numeric values in XML/YAML were emitted/parsed incorrectly
695 on some non-US locales in Linux
696 * VS2005 Express/Standard Editions could not build OpenCV
697 (due to missing OpenMP support)
701 * Certain tests may fail on IA64 (Itanium), e.g.
702 gaussian filter, fundamental matrix estimation, chessboard detector.