Yes, center of gravity is the centroid.
By the way, this algorithm assumes that images are very similar (in contents and brightness). So, it is not designed to register images of an eclipse at several stages, a transit, or other variable phenomena. The reason is that the main function that we are optimizing is the mean quadratic error between the images. There are other functionals that work much better with highly different images, like those you may find in medical imaging, for example (in MRI, depending on the sequence and parameters, you may end with two very different images, of the same field).
