初学，图像识别什么都不懂，一点点摸索
环境：win2008 r2 x64, vs2008 日常x86编译
下载了最新的opencv 3.2.0
发现从3.0开始，opencv只自带x64的链接库，没有x86，只好自己用开源代码重新编译x86版本
摸索了几个小时，发现因为底层库的关系，已经不再支持VS2008，哪怕重新编译OPENCV源代码
最终选择2.4.13，最后一个能支持VS2008的版本。
文档地址：http://docs.opencv.org/2.4.13/
解压缩到d:\opencv，因为更深的目录，不支持中文，他默认编码集是unicode
使用cmake，
源地址：D:/opencv/sources
目标地址：D:/opencv/build_byFBI
configure：VS 9 2008
然后一路测试，把有问题的组件都去掉（一般是doc，CUDA），等都没问题了，就Generate（生成sln）
然后用vs2008打开，全部重编译。
1）在Debug下，选择(解决方案资源管理器)里的 Solution OpenCV（解决方案“OpenCV”），点右键，运行"Rebuild Solution"；如编译无错误，再选择INSTALL项目，运行"Build"。
2）在Release下，同理
重编译要得到的东西：Debug版本和Release版本的dll，lib，头文件，都在install文件夹里，其余文件可以删除。
在项目的配置中，加上【附件包含目录】"..\opencv\opencv-2.4.13\build\include"
代码中加上：
#ifdef _DEBUG
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_calib3d2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_contrib2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_core2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_features2d2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_flann2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_gpu2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_highgui2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_imgproc2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_legacy2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ml2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_nonfree2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_objdetect2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ocl2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_photo2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_stitching2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_superres2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ts2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_video2413d.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_videostab2413d.lib")
#else
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_calib3d2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_contrib2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_core2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_features2d2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_flann2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_gpu2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_highgui2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_imgproc2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_legacy2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ml2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_nonfree2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_objdetect2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ocl2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_photo2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_stitching2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_superres2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_ts2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_video2413.lib")
#pragma comment(lib,"../opencv/opencv-2.4.13/build_byFBI/install/x86/vc9/lib/opencv_videostab2413.lib")
#endif
另外，这后面是准备素材：
在cmake官网下载cmake-3.9.1-win64-x64.msi，https://cmake.org/download/
我不用python，所以就不下载Python libraries、Numpy库了
在intel官网下载Intel Threading Building Blocks (TBB)，https://software.intel.com/en-us/intel-ipp，在opencv官方介绍是说Using this will make sure that the OpenCV library will take advantage of all the cores you have in your system's CPU.
在eigen官网下载Eigen 3.3.4，他是一个线性代数的C ++模板库，http://eigen.tuxfamily.org/index.php?title=Main_Page#Download
如果你有NVIDIA显卡，强烈建议下载安装CUDA安装包，http://developer.nvidia.com/cuda-downloads，他将允许您使用GPU内部的电源。这将大大提高某些算法的性能。
OpenNI框架下载，https://structure.io/openni
Doxygen下载，http://www.stack.nl/~dimitri/doxygen/download.html
不用qt框架，所以不装了。
自此，素材下载全了。天哪，这么多
参考文献：http://blog.csdn.net/abcCL6/article/details/43061349
感谢，给了很大启发。