OpenCV 1.1pre1编译方案及其在VisualC++6.0的配置

因库函数支持问题，今天卸载了OpenCV1.0 ，安装了OpenCV_1.1pre1a ，并把我的项目在这个新的环境重新编译了，以后就暂时用OpenCV_1.1pre1a 支持库了。

在配置OpenCV_1.1pre1a环境时遇到不少问题，网友发言很热烈，但总没找到一个能完全解决这些问题的发言。在这里把自己在VC++6.0环境下成功使用OpenCV_1.1pre1a的过程写下来，以便于交流。

1、先来看下OpenCV_1.1pre1a的ChangeLog(摘自OpenCV安装路径下的说明文档)

=============================================================================
                      OpenCV 1.1pre1, October, 2008
=============================================================================

>>> New functionality/features: <<<

  - General:
    * Octave bindings have been added. See interfaces/swig/octave (for now, Linux only)
    * [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).
    * [Windows] Python bindings have been updated to use Python 2.6
    * [Windows] cvcam has been removed (as videoInput is now supported by highgui)

  - CXCORE, CV, CVAUX:
    * Speeded-up Robust Features (SURF), contributed by Liu Liu.
      see samples/c/find_obj.cpp and the documentation opencvref_cv.htm
    * Many improvements in camera calibration:
      - Added stereo camera calibration: cvStereoCalibrate, cvStereoRectify etc.
      - Single camera calibration now uses Levenberg-Marquardt method and supports
        extra flags to switch on/off optimization of individual camera parameters
      - The optional 3rd radial distortion parameter (k3*r^6) is now supported in
        every calibration-related function
    * 2 stereo correspondence algorithms:
      - very fast block matching method by Kurt Konolige
        (processes the Tsukuba stereo pair in <10ms on Core2Duo laptop)
      - slow but more accurate graph-cut based algorithm by Kolmogorov and Zabin
    * Better homography estimation algorithms (RANSAC and LMEDs)
    * new C++ template image classes contributed by Daniel Filip (Google inc.).
      see opencv/cxcore/include/cvwimage.h
    * Fast approximate nearest neighbor search (by Xavier Delacour)
    * Codebook method for background/foreground segmentation (by Gary Bradski)
    * Sort function (contributed by Shiqi Yu)
    * [OpenCV+IPP] Face Detection (cvHaarDetectObjects) now runs much faster
      (up to 2x faster) when using IPP 5.3 or higher.
    * Much faster (~4x faster) fixed-point variant of cvRemap has been added

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

  - HighGUI:
    * [Windows, 32bit] Added support for videoInput library.
      Hence, cvcam is [almost] not needed anymore
    * [Windows, 32bit] FFMPEG can now be used for video decoding/encoding
      via ffopencv*.dll
    * [Linux] Added unicap support
    * Improved internal video capturing and video encoding APIs

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

  - New samples (opencv/samples):
    * Many Octave samples
    * find_obj.cpp (SURF), bgfg_codebook.cpp (Codebook BG/FG segmentation),
      stereo_calib.cpp (Stereo calibration and stereo correspondence)

>>> Bug fixes: <<<
     Many thanks to everybody who submitted bug reports and/or provided the patches!

    * added dma_unlisten to icvCloseCAM_DC1394 (thanks to Victor Benso)
    * LMEDs algorithm for cvFindFundamentalMat estimation has been fixed
    * Broken debug build of highgui in VS2005/2008 (SF #2056185, SF #1709435)
    * cvFindChessboardCorners memory leak and incorrect size checks
      (SF #1972896, SF #1910323)
    * disabling GTK causes v4l runtime error (SF #2088102)
    * cvsetmousecallback bug (SF #2053529)
    * libhighgui needed deprecated "img_convert" replacement (SF #2011753)
    * Segfault in createsamples caused by uninitialized variable (SF #1977989)
    * Data Alignment Issue in bgfg_gaussmix (SF #1961755)
    * libpng need to be updated (SF #1952793)
    * cvCreateVideoWriter_Win32 - identifier not found (SF #1944254)
    * Bug in cvmorph.cpp (SF #1908844)
    * dilate (cvDilate) works bogus with default kernel (SF #1887130)
    * CvEM non-default constructor declared but not defined (SF #1830346)
    * cvFloodFill (in ver 1.0) Hangs (SF #1790565)
    * double delete in CvImage (SF #1733437)
    * cvFilter2D anchor default value is not working properly (SF #1713779)
    * cvCvtColor - Bug? in converting HSV2RGB (SF #1676344)
    * Invalid selection of the MKL-dll version in cvUseOptimized() (SF #1633017)
    * ...

>>> Known issues:
    * Borland compiler is not supported (but might work) in this update.
    * 64-bit Windows is not supported (but might work) in this update.
    * SF bug tracker still contains over 50 records of open bugs.
      Many of them will be addressed in the next update.

2、安装 OpenCV_1.1pre1a

OpenCV_1.1pre1a 官方下载：
http://sourceforge.net/projects/opencvlibrary/files/opencv-win/1.1pre1/
OpenCV_1.1pre1a 安装基本同 OpenCV 1.0 安装
可以参考：
http://www.opencv.org.cn/index.php/VC6%E4%B8%8B%E5%AE%89%E8%A3%85%E4%B8%8E%E9%85%8D%E7%BD%AE

3、编译 OpenCV_1.1pre1a
在你的OpenCV安装目录下找到_make文件夹，用Visual C++ 6.0打开工作空间 opencv.dsw 对这里的每一个工程进行逐个编译，逐个编译时并不是按工程从上到下编译，而是有先后顺序的，如你编译工程 highgui 会提示连接错误，不能连接libjpegd.lib，这时你需要先编译工程libjpeg，其他工程编译类似处理。

在编译工程cvaux，还会遇到一个小小的错误

error C2039: 'foreground_regions' : is not a member of 'CvFGDStatModel
处理如下：
搜索关键字foreground_regions，
第1137行
CvMemStorage* storage; /*storage for 揻oreground_regions?/ /
CvSeq* foreground_regions /*foreground object contours*/
改为如下形式：
CvMemStorage* storage; /*storage for foreground_regions*/ /
CvSeq* foreground_regions /*foreground object contours*/

每一行后面的这些"/"是表下一行是上一行的接续.如
define可以替代多行的代码，例如MFC中的 非常经典的 宏定义
#define MACRO(arg1, arg2) do { /
/* declarations */ /
stmt1; /
stmt2; /
/* ... */ /
} while(0) /* (no trailing ; ) */
关键是要在每一个换行的时候加上一个"/"

编译好所有的工程后我们就可以配置环境了

4、VC6下安装与配置
OpenCV_1.1pre1a 在VC6下安装与配置过程基本同 OpenCV 1.0
可以参考：
http://www.opencv.org.cn/index.php/VC6%E4%B8%8B%E5%AE%89%E8%A3%85%E4%B8%8E%E9%85%8D%E7%BD%AE

不同的地方是：
在 配置Visual C++ 6.0 时
全局设置 因为在OpenCV_1.1pre1a，"cvcam has been removed" ，所以相应的，所有关于cvcam的配置项目我们不需要配置。

项目设置 变为：
每创建一个将要使用OpenCV的VC Project，都需要给它指定需要的lib。菜单：Project->Settings，然后将Setting for选为All Configurations，然后选择右边的link标签，在Object/library modules附加上

cvd.lib cvauxd.lib cxcored.lib highguid.lib mld.lib cxtsd.lib cvhaartraining.lib

如果你不需要这么多lib，你可以只添加你需要的lib。

另 ：在 C:/Program Files/OpenCV/otherlibs目录下有个 _graphics 文件夹("This folder contains libraries and headers of a few very popular still image codecs used by highgui.")，其下有include、src、lib三个文件夹，如有需要，我们可以参照前面的相应配置进行。

到此VC6下OpenCV_1.1pre1a安装与配置基本完成， 我们可以 运行opencv.dsw里的工程cvtest，看看一些函数的测试结果。

完成以后步骤后，我就开始重新编译我的项目了，OK ，一切正常！