How does findChessboardCorners() find its corners

edit

I am attempting to detect right-angle corners in an image. I notice that cv::findChessboardCorners() does exactly this and really well. So I'm interested to know how it works. I've googled around but cannot seem to find a simple explanation of what steps it takes.

I did find a link to the code so I can see that it performs some smart thresholding (which I don't quite understand what its doing) but from there its hard to figure out what its doing. Does it use a corner detector at any point? How does it distinguish right-angle corners from any other corner?

Is anyone able to explain simply what algorithms/techniques its using?