Inverse 2D integration

[CiolJegou]

Hello,
I am trying to go back from a 2D integrated image to the original image and I am encountering some issues.
My naïve approach is to integrate a "column" image and a "row" image (with the same poni file and integration parameters), thinking that I can use these results to map back any integrated images. However it is not the case and after many attempts I do not understand why.

Do you know where am I doing things wrong ? And maybe is there already a method to perform 2d "inverse integration" in pyFAI ?
Thank you !

[kif]

I am not sure this can be made the way you are approaching the problem: each bin not only mixes several pixel but each pixel can even be split over several bins as well.

Why don't you use the calcfrom2d method of the azimuthal integrator ? it is based on bilinear interpolation but the result is usually pretty good.
http://www.silx.org/doc/pyFAI/dev/api/geometry/index.html#pyFAI.geometry.core.Geometry.calcfrom2d

[kif]

There is an example of application in this tutorial (almost at the end):
http://www.silx.org/doc/pyFAI/dev/usage/tutorial/symmetrize.html