Well, If this is a blog on some of the cool stuff in Computer vision, I think i must mention about the api’s that one can use, or how to setup your system.
So yes, there is a myraid choice available to start developing. I personally like OpenCV because of its’ enormous community support along with its multiplatform facilities.
Apart from this, there is MATLAB. Whoaaa… this is like a computing powerhouse. With so many easy to use functionality, and simply clicking your way into outputs, MATLAB has one hell of a pricetag.
Apart from these, there are :
- SimpleCV : This is actually computer vision made easy.There is no need to spend endless time in figuring out data structures, Matrices, what color format to select ( 8UC1, 32FC1,etc)
- VxL Libraries (I like them too.. I’ll use them the day I make a robot :P )
- VLFeat : This may come out very handy !! It contains some of the finest implementations in vision as well as image understanding algorithms
- ImageJ : A java library for primarily image processing
- CCV ( Haven’t used it yet, so .. what do i know )
Anyway, moving on to the installation, there are plenty of online resources that can be used for downloading and setting up OpenCV.
For compiling opencv c++:
- Use CMAKE. It’s much simpler to compile using cmake
You can use this boiler plate file if incase .
# Filename : CMakeLists.txt
cmake_minimum_required(VERSION 2.8)
project( project_name )
find_package( OpenCV REQUIRED )
add_executable( project_name project_name.cpp )
target_link_libraries( project_name ${OpenCV_LIBS} )
simply execute :
cmake .
make
./project_name
Incase you’re using python,
python filename.py <args>
To be continued