Hadoop上结合opencv\javacv

fdfe

　　mac上安装opencv
　　1. 去 http://opencv.org 下载最新版OpenCV for Linux/Mac源文件，目前版本是2.4.3。下载后解压。
2. 去 http://www.cmake.org 下载最新版cmake，.dmg文件，目前版本是2.8.10.2。下载后安装。
3. 接下来开始安装opencv
    开一个终端，进入opencv解压缩的目录，如cd 然后把文件夹拽进去。
    mkdir release
    cd ./release
    cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D BUILD_PYTHON_SUPPORT=ON -D BUILD_EXAMPLES=ON ..
    make
    sudo make install
4. 至此opencv应该已经装上了。
        注：如果cmake出现错误，请到苹果开发者网站那里下载Command Line Tools安装。（关键）
　　在MacOS和iOS系统中使用OpenCV
　　http://blog.devtang.com/blog/2012/10/27/use-opencv-in-ios/
　　在Xcode使用openCV
　　http://blog.iyunv.com/eminia/article/details/6635117
　　JavaCV的配置及使用
　　http://blog.iyunv.com/victory06057231/article/details/8026188
　　
Hadoop Streaming does nto really agree with binary data, so I wouldn't go there.
You will have to write both a custom Input/OutputFormat in Java plus the face-detection code in C++ (see http://wiki.apache.org/hadoop/HadoopStreaming )
Why not use OpenCv Java bindings? ( http://docs.opencv.org/java/ ,http://docs.opencv.org/doc/tutorials/introduction/desktop_java/java_dev_intro.html )


opencv如何从内存读取图像数据？
如果你的每一帧数据都能得到
IplImage* img=cvCreateImageHeader(cvSize(1920，1080),IPL_DEPTH_8U,3);//3表示24位，灰度请改为1
img->imageData  = 你的帧内存数据;
接下面你可以处理这一帧图像了

http://code.google.com/p/javacv/issues/detail?id=116

Any idea on how to write an openCV programas a MapReduce job ?
http://www.linkedin.com/groups/Any-idea-on-how-write-3990648.S.234687769

Amazon Elastic MapReduce
http://www.atlantbh.com/amazon-elastic-mapreduce-part-1/

mapreduce 与 javacv
http://www.atlantbh.com/category/big-data/hadoop-as-service/

Hadoop Streaming高级编程
http://dongxicheng.org/mapreduce/hadoop-streaming-advanced-programming/

javacv可读的hadoop inputformat
https://code.google.com/p/hadoop-computer-vision/source/browse/trunk/hadoop-computer-vision/src/?r=20#src%2Fedu%2Fvt%2Finput