[Feature request] ML type data processing

[borondics]

This feature request is for the near future but not in the coming weeks.

It would be great to have the ability of doing processing (learning) on a subset of an image and then applying this knowledge on the whole image. I am afraid that this ability is what we lost with the solution we handle the 2D images?

The use case is actually awesome and once again, we are breaking ground here as no other software really is able to do such things. Imagine that you have an image, which you select some values from (either by hand or by using a Distributions widget histogram), then fit a background fit on that and you apply the correction to the whole image to make sure that the calculation avoids the edges of some surface features and thus you get no funny data at the end.