We have 2 ways in order to install OpenCV on ours Ubuntu machine.
Here I explain both of them:

  1. from repositories
  2. from sources
  3. changes for Ubuntu 11.10

Then we can check if the python wrapper is properly installed.



INSTALL FROM REPOSITORIES

Prerequisites

  • C++ compiler like g++ (sudo apt-get install g++)
  • GTK+ 2.0 or higher (sudo apt-get install libgtk2.0-dev)


Procedure

  1. Open a terminal and type “sudo apt-get install” for all these packages 
    libcv2.1
libcv-dev
libcvaux2.1
libcvaux-dev
libhighgui2.1
libhighgui-dev
opencv-doc
python-opencv
  2. Then, always in a terminal, type: 
    export LD_LIBRARY_PATH=/home/opencv/lib
export PKG_CONFIG_PATH=/home/opencv/lib/pkgconfig

    The above ones are default paths for the opencv libraries.

  3. To check the path where opencv & other lib files are stored (needed to compile your opencv programs) type (and check the output): 
    $ pkg-config --cflags opencv
         -I/usr/include/opencv

$ pkg-config --libs opencv
         -lcxcore -lcv -lhighgui -lcvaux -lml
  4. Create a test file “hello.c” and type this code in it: 
    #include <stdio.h>
#include "cv.h"
#include "highgui.h"

int main( int argc, char** argv )
{
    /* data structure for the image */
    IplImage *img = 0;

    /* check for supplied argument */
    if( argc < 2 ) {
        fprintf( stderr, "Usage: loadimg <filename>\n" );
        return 1;
    }

    /* load the image,
    use CV_LOAD_IMAGE_GRAYSCALE to load the image in grayscale */
    img = cvLoadImage( argv[1], CV_LOAD_IMAGE_COLOR );

    /* always check */
    if( img == 0 ) {
        fprintf( stderr, "Cannot load file %s!\n", argv[1] );
        return 1;
    }

    /* create a window */
    cvNamedWindow( "image", CV_WINDOW_AUTOSIZE );

    /* display the image */
    cvShowImage( "image", img );

    /* wait until user press a key */
    cvWaitKey(0);

    /* free memory */
    cvDestroyWindow( "image" );
    cvReleaseImage( &img );

    return 0;
}
  5. Compile it 
    g++ -I/usr/include/opencv -lcxcore -lhighgui -lm hello.c -o hello
  6. Then run it 
    ./hello img.jpg

    where “img” is the name of any image within the same folder of the C script.
    If this not run, try

    export PATH=$HOME/usr/bin/:$PATH

    and go to step2 again


  7. In order to simplify the command used for compile the script, we can create a shortcut
    • go to your local home directory: 
      cd /home/
    • open the .bashrc file (it will be hidden): 
      vi .bashrc
    • append this line: 
      alias gcv="g++ -I/usr/include/opencv -lcv -lcxcore -lcvaux -lhighgui -lm"
    • save and clode the file, then close the terminal and open it again
  8. Now go to directory containing a sample program and type: 
    gcv hello.c -o hello
./hello img.jpg



INSTALL FROM SOURCE

Further prerequisite: CMake 2.6 or later

Procedure

  1. Get the latest stable OpenCV version downloading the source tarball (OpenCV-x.y.z.tar.bz2)
  2. Create a folder named “opencv” (for example in the folder “/projects”) and extract in it the content of the tarball
  3. Then open a terminal and type: 
    cd ~/projects/opencv # the directory containing INSTALL, CMakeLists.txt etc.
mkdir release
cd release
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D BUILD_PYTHON_SUPPORT=ON ..
  4. Then, build the library and install it to the target directory: 
    make
sudo make install



CHANGES FOR UBUNTU 11.10

If you get some library errors, like the following:

./hello: error while loading shared libraries: libcv.so.2.1: wrong ELF class: ELFCLASS64

you can solve the problem typing:

export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH
sudo ldconfig

and then recompiling the script in this way:

g++ hello.c -o hello `pkg-config --cflags --libs opencv`



CHECK THE PYTHON WRAPPER

In order to check if the python wrapper is properly installed create a new file named “hello_python.py”

import cv
im = cv.LoadImage( "sample_image.jpg" )
cv.ShowImage( "sample1", im )
cv.WaitKey()

Then open a terminal and type:

python hello_python.py