/connect-component-analysis

Connected Component Labelling

Implementation of Connected Component Labelling Algorithm explained in http://aishack.in/tutorials/connected-component-labelling/ using OpenCV

Prerequites

  • git
  • CMake >= 2.8
  • OpenCV >= 2.4
  • C++ 11 support
  • Visual Studio 2015 (for Windows users)
  • boost
  • mysqlcpp
  • fann

Configuration Instruction

Windows user

For windows user, please use CMakeLists_Windows.txt before running CMake to compile it. Steps

  1. Rename CMakeLists_windows.txt to CMakeLists.txt

  2. Ensure that you have already installed OpenCV on your Windows operating system. And also you have already linked to system path the bin directory of OpenCV.

  3. Change variable OpenCV_dir inside CMakeLists.txt to conform with your OpenCV build directory

  4. Ensure you have already included cmake executable in system path, so you are able to use it in PowerShell or CMD. (I suggest PowerShell)

  5. Open your power shell, clone this repository by issuing

    git clone https://github.com/kennykarnama/connected_component_labelling.git

  6. Once you done cloning it, just issue command inside the repository directory

cmake .

  1. If it has finished compiling then open connected_components.sln using your visual studio editor.
  2. Head to Build --> Rebuild ccl.
  3. After completing, head to debug directory, you will find ccl.exe
  4. Done :D

Linux user

For Linux user, please use CMakeLists_Linux.txt before running CMake to compile it.

  1. Rename CMakeLists_Linux.txt to CMakeLists.txt
  2. Ensure you have already installed OpenCV correctly.
  3. Just run cmake .
  4. After CMake compiled successfully, build the executable using make
  5. You are all set now

Running the program

For executing this program, below are the pattern to run it :

./ccl [image path file] [dummy args]

  • [image path file] --> path to your picture files (ensure it is in binary or black-white picture)
  • [dummy args] --> just past whatever string to this second argument (initially it was meant to support optional arguments, but not finished yet)

Demo

Alt Text

Contributions

Any contributions will be appreciated because this program is still not ideal in design pattern, algorithm efficiency and code structuring. It also lacks of tests !

License

MIT