Pairwise CC is too slow and heavy to compute

[ekatrukha]

For example, for 120 image volumes, one needs to compute $(N * 2 - N)/2 = 0.5 * N * (N-1) =7140$
two volumes cross-correlations. That is a lot and very time-consuming for 3D data.
It is even more heavy, since we are using masked cross-correlation.
Current masked cross-correlation implementation lives in the GenNormCC.java class.
Some time ago it was written as a proof-of-principle and is not optimized
for a fast performance.

Why the masked cross-correlation is slower to compute?

Again, it is slower to compute in comparison to what? Let's consider some alternatives and provide some estimates.
TODO