Hi Oori,
It should be sufficient to apply a mask in SurfSmooth. I have never
tried that out, but if the smoothness is indeed evaluated and applied
within the mask, that would be all you need (along with actually masking
the random noise, though I assume you do that).
SurfClust could use a mask option as well, but it is really not necessary
in this case.
I was going to mention slow_surf_clustsim.py as Peter did. Do you know
how your implementation compares with it?
Note that slow_surf_clustsim.py does not have a mask option, but I could
look into that if it would be useful.
- rick
