Opencv2.4.3 freshly baked

Opencv2.4.3 has finally been released, and it has been four months since the last release, which is a big improvement compared to the previous release.

For: OpenCV-2.4.3.exe

 

Opencv2.4.3 adds many new features and becomes more Nb, the main reference: http://code.opencv.org/projects/opencv/wiki/ChangeLog#243

Many of the changes in this version come from Google summer code (gsoc), and that project contributed a lot of code.

 

• Multithreading
• Background added many general multi-thread parallel processing algorithms: TBB (cross-platform: http://threadingbuildingblocks.org), OpenMP (cross-platform: http://en.wikipedia.org/wiki/OpenMP ),
  Grand Central Dispatch (IOS/MacOSX Apple platform: http://en.wikipedia.org/wiki/Grand_Central_Dispatch), concurrency
  (Visual Studio 2010 or later Microsoft: http://msdn.microsoft.com/en-us/library/dd492418.aspx), c = (http://www.hoopoesnest.com/cstripes/cstripes-details.htm ).
  Many existing parallel opencv algorithms are embedded with these algorithms. Therefore, you can use them without installing the libraries TBB.

• Android platform
• Added opencv management and a new Java sample framework to better support cameras on Android. More details: Android Release Notes
  .

 

• IOS
• Opencv2.frameworkIt is compatible with ios6-And iphone5-(armv7s. thanks to many important algorithms (e.g. face detection, bilateral filter, etc .) it can run more efficiently on a5 or dual-core processors. We also fixed the bug of building a script framework. Now we use the optimized "-O3" framework to replace the "-O0 framework" on opencv 2.4.2 ".
• Finally, our gsoc student Eduard feicho and Charu Hans has written detailed instructions on how to install opencv in IOS apps. Please, check: http://docs.opencv.org/doc/tutorials/ios/table_of_content_ios/table_of_content_ios.html.

• Python
• Some of our other gsoc students, Alexander mordvintsev, added the python sample, including asift, hand-written Digit Recognition, motion debluring and even the use of Amazon pickers. (asift, hand writing recognition, motion denoising and Amazon picloud usage) checkhttps: // github.com/itseez/opencv/tree/master/samples/python2

• Cascade Detector
• Gsoc student Attila Novak has trained two new cascade classifiers to detect silverware ?) And side face, checkhttps: // github.com/itseez/opencv/tree/master/data/lbpcascades

• Image Denoising,
• Another result from gsoc (that was a good year !) : Victor passichenko has improved the non-local means algorithm. Although this algorithm is not real-time, the effect is quite good! Please, check: http://docs.opencv.org/trunk/modules/photo/doc/denoising.html

• Optical Flow
• Yet another result from gsoc (that was a good year indeed !) : Yuri zemlyansky added a "simple flow" optical flow algorithm; check: https://github.com/Itseez/opencv/tree/master/samples/cpp/simpleflow_demo.cpp.

• Many advances and new features in GPU models:
  -Device layer opened for users; now users write Cuda device code using opencv. Driver layer for users: users can write Cuda drivers using opencv;
  -In cascade support in cascade classifier, the cascade classifier supports the cascade;
  -Fast non local means image denoising is added with a fast non-local mean image denoising algorithm;
  -Faster integral image Calculation on Kepler introduces faster image credit Calculation on Kepler;
  -Hough circles, Hough lines and Generalized HOUGH Transform implementation huff circle, HUF straight line and generalized HUF transformation;
  -Bilateral filter: Bilateral Filtering;
  -Generalized flood fill connected component labeling: Fill in the Connected Domain annotation by using the generalized water filling method;
  -Background/Foreground Segmentation Algorithms: mixture of Gaussian, vibe, GMG (. godbehere,. matsukawa, K. goldberg ). foreground background separation algorithm, addition of GMM, vibe, and GMG Algorithms
  -Added confidence calculation into hog: The Trust Computing function is added to the hog class.

• In opencl, arithmetic operations, filtering, common geometric transformation, optical flow, and cascading classifier are added. Currently, these algorithms can only run on the GPU.

• Based on opencv2.4.2, about 130 bugs are fixed.

• Opencv is now widely promoted and has been added to GitHub. The URL is:Http://github.com/itseez/opencv. You can download the latest version on this page and submit various questions ..

• Since 2.4.3rc fixed some problems. encv-based applications on xcode 4.5 & ios6, opencv + TBB on Windows etc.) Some problems were also disabled. For details, refer to: code.opencv.org.

 

 

Let's review some new features of opencv2.4.2 (the platform is not mentioned, but functional functions ):

• EPFL group (kirell Benzi, Raphael Ortiz, Alexander re alahi and Pierre vandregheynst) contributes a new feature description operator freak: It is said that the second kill orb and surf, and it is faster than orb, please, seesource: // trunk/opencv/samples/CPP/freak_demo.cpp.

• Excellent facial recognition improvement tutorial added by Philipp Wagner. Check theface
  Recognition tutorial.

Refer:

Http://code.opencv.org/projects/opencv/wiki/ChangeLog

Http://www.cnblogs.com/tornadomeet/archive/2012/11/07/2759476.html