Process image coming from a OpenGL rendering context

edit

Hi all! I have an image coming from an OpenGL rendering context:

We have alpha background both inside and outside the black contour. I'd like to know whether the image inside the black contour contains alpha values. If yes, I'll do some processing in the scene in order to reduce the alpha (basically scaling the image sprite), then export the render again and do the same process, until we have no alpha pixels inside the contour. I think that OpenCV is right technology to do the image scanning. I have done some basic stuff before with OpenCV, like altering the colors of each pixel of an image, but I think I need more experience this time. Do you know which could be the best approach? We are using OpenCV 2.4.7 in a C++ project.

Thanks, Robert.