2127 flat, dark, background in pyFAI-calib2 [PROPOSAL]

[EdgarGF93]

Proposal for including dark current and flat field images in pyFAI-calib2

• Dark current file(s)
• Flat field file

Idea:

• Currently, pyFAI-calib2 works with ImageModels to handle the data arrays from the file and the mask. ImageFromFilenameModel is a generic class that handles the filename and the data array. MaskedImageModel computes the masking of the array.
• Idea: to include two new classes:

1. DarkedMaskedImageModel, same as MaskedImageModel but also takes an instance of ImageFromFilenameModel for the dark current file, and computes the masking and following subtraction of the dark current array

darked_masked_data = Masking(data_array) - dark_current_array

2. CorrectedImageModel, takes two instances of DarkedMaskedImageModel and an instance of ImageFromFilenameModel for the flat field file, and computes the subtraction of both DarkedMaskedImageModel following by a normalization with flat field array. The idea is to use the second instance of DarkedMaskedImageModel as a background file. This is the correction protocol:

corrected_image = (darked_masked_image_data - darked_masked_background_data) / flat_field

inspired by: https://lookingatnothing.com/index.php/archives/2251

[EdgarGF93]

Now there is an instance of a PreProcessedImageModel, which takes into account the mask, the flat and the dark
The flat and the dark now can be uploaded through CLI and GUI
In principle, the preprocessed image now appears during the PeakPicking and Geometry tasks, and flat and dark are used to integrate in 2d and 1d in the final task (IntegrationTask)

Most likely, there are still things to wire if the image is changed during the tasks

[kif]

Can't this be replaced by preprocessedImageModel ?

[EdgarGF93]

correct, DarkedMaskedImageModel.py is no longer needed

[kif]

Is this still WIP ?

[EdgarGF93]

Yes, I will test the change of the image and I will label Ready to merge

[EdgarGF93]

now, flat and dark images can be uploaded through CLI, dialog file and drag&drop, and they will be properly computed along all the calib2 tasks

[kif]

@jonwright are you willing to test it ?

[kif]

LGTM

[kif]

Is this cached anywhere ?

[jonwright]

Sure - we will try this later today or during the weekend - sometime soon after we figure out how to install pyFAI from a pull request.

[kif]

you can use `./bootstrap pyFAI-calib2` to test locally without installing. You may install all dependencies before-hand with a a "pip install -r requirements.txt --upgrade"