My question is on the registrate 2 points cloud .
There are some  examples in the vxl/rpl/rgrl package to demostrate Range
data registrate algorithms,but there is only rgrl-trans-rigid that
without any scale between 2 data sets.

My data sets exist a little scale difference,so i need a 3D similarity
transform(translate + rotate +scale).But i haven't find such a
transform class in vxl/rpl/rgrl.
How should i do? 
vxl/rpl/rgrl also provide a rgrl_est_similarity2d but it only works in 2D.
If I use a 3D affine estimate instead,it will be over-estimate?

Thank you