C++ Development Environment Setup for Vision, Image Processing, Graphics & Parallel Programming

Programming, Tutorial

C++ is an amazing programming language and the recent updates in STL with C++11 & C++14 make it even more handy. However, there are a lot of programmers interested in using C++ that end up losing interest due to lack of a decent setup for small programs. In this post, I am going to provide a very simple yet stable development process that provides a way to create portable applications easily.

We will be using a combination of cmake, gcc, a bunch of common libraries, CUDA & git.

All these are quite used and there are very many tutorials about setting the environments in different operating systems. We will look at each of the steps of the dev process individually.

C++ and libraries

If you take an example as following, modified to work with OpenCV3+ :

#include <opencv2/highgui.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/imgcodecs.hpp>

using namespace cv;

int main( int argc, char** argv )
{
		char* imageName = argv[1];
		Mat image;
		image = imread( imageName, 1 );
		if( argc != 2 || !image.data )
		{
				printf( " No image data \n " );
				return -1;
		}

		Mat gray_image;
		cvtColor( image, gray_image, CV_BGR2GRAY );
		imwrite( "Gray_Image.jpg", gray_image );
		namedWindow( imageName, CV_WINDOW_AUTOSIZE );
		namedWindow( "Gray image", CV_WINDOW_AUTOSIZE );
		imshow( imageName, image );
		imshow( "Gray image", gray_image );
		waitKey(0);
		return 0;
}

Here we use opencv to read a color image, convert it into gray-scale, save it in a file and also display both the images. For this sort of code to be able to run succesfully, you need the following:

The tutorials on how to get them up and running are available widely. All you need to do is a default installation without any tricks.

Building code

Once the code is written, it needs to be compiled and run as an executable. Different operating systems require executables in different ways and ‘cmake’ provides a nice cross-platform way to build these executables from the same source code.

All you need is a simple CMake template that looks the following way:

cmake_minimum_required(VERSION 2.8)
project( MyCVProject )

option( WITH_CUDA "CUDA Enabled" OFF)

set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")

find_package( OpenCV REQUIRED )
find_package( CUDA )

include_directories(${CMAKE_CURRENT_SOURCE_DIR})
include_directories(${OpenCV_INCLUDE_DIRS})


if(WITH_CUDA)
  if(NOT CUDA_FOUND)
    message(FATAL_ERROR "Please install CUDA and make it available for Cmake
      before compiling the code with the WITH_CUDA flag")
  endif ()
endif ()

if(WITH_CUDA)
  cuda_add_executable( DisplayImage DisplayImage.cpp)
  target_link_libraries(DisplayImage ${OpenCV_LIBS})
else ()
  add_executable( DisplayImage DisplayImage.cpp)
  target_link_libraries( DisplayImage ${OpenCV_LIBS})
endif ()

This needs to be placed in a file name ‘CMakeLists.txt’ along with the source. Using the cmake system, we can generate further build files that are dependent on the OS.

Then, you can use the build files to further build using the compilers mentioned above to make the executable required to run the program. The main advantage of using cmake system is that all the dependencies on libraries are resolved using the default installations. This is the main reason that the code becomes portable and easy to distribute.

Versioning of code

A lot of programming works by developing through stages. It is definitely worth it to save these stages as they are. This is where a versioning system comes very handy. There are several possibilites, but ‘git’ is my favorite. Also, Github provides an amazing free service if you don’t mind keeping the code opensource.

Once you create a repository on github, you can follow the simple instructions there to set it up on your local machine. A lot of tutorials are available for getting a good overview of how git can be used for a big project. A really good one is available here.

In simple words, this is the cycle of staging stuff to git.

1. Make code changes
2. Add them to git using git add
3. Write a commit message for this staging git commit
4. Push them to remote git push origin master
5. Go back to step 1.

In addition, you can sync up your local repository by doing git pull. That’s it really!

Organisation of source files

It is really important not to add binary files to the git, unless they cannot be avoided somehow. Especially the files that need to be ignored are the executables that can be built using the source files. You can use the .gitignore in the repository to add a list of extensions or files to be ignored.

In addition, the advantage of using cmake is that you do not have to be in the same folder as your source code. You can make your changes in the source folder, then build and test in an entirely different folder without having to copy anything. All you need to do is call cmake $source_dir from your build directory. This helps in keeping the source directory clean.

In order to get started, I am providing an empty template here to use. It includes the necessary files to start programming and I hope you have fun with it!