compute operation Normalising

[ashmeigh]

Issue

This issue was part of #2051

Description

This pull request addresses the ongoing effort to standardize the execution of operations by transitioning from the partial function style to a structured compute_function approach. In this update, several operations have been refactored to adhere to the new standard. The modifications ensure a clearer separation between the setup and execution phases of each operation, improving the maintainability and readability of the codebase.

Changes

Refactored multiple operations to implement the new compute_function approach.
Introduced static methods compute_function in each affected operation module to encapsulate the core logic for executing the operation on a per-slice basis.
Updated the calling code to use the compute_function method instead of the previous partial function style

[coveralls]

coverage: 72.685% (-0.05%) from 72.735%
when pulling 204d1f1 on compute_norm
into 619bdd4 on main.

[samtygier-stfc]

Here are the notes from our catch up.

Neutron flux is variable, so each image has different exposure. The aim of the algorithm is to make the mean air region in each slice equal to each other, by multiplying each frame by a normalisation factor. So for both cases the final step is to have a normalisation array the length of the number of slices, and multiple each slice by the corresponding value.

For the Stack Average case, we take the average of the air region for each slice to get the normalisation array. Then this itself needs normalising to its own mean, before applying to the full image stack.
A non-parallel version would look like:

air_means = images.data[:, air_top:air_bottom, air_left:air_right].mean(axis=1,2)
#[50,48,52,49, ...]
normed_air_means = air_means / air_means.mean()
#[1,0.96,1.04, ...]

for i in range(images.data.shape[0]):
    images.data[i] /= normed_air_means[i]

# so if you looked at the means again, they would all be the same value
new_means = images.data[:, air_top:air_bottom, air_left:air_right].mean(axis=1,2)
# [50,50,50,50, ...]   

For the Flat Field mode, we normalise to the flat field. e.g.

air_means = images.data[:, air_top:air_bottom, air_left:air_right].mean(axis=1,2)
#[50,48,52,49, ...]
flat_means = flat_field.data[:, air_top:air_bottom, air_left:air_right].mean(axis=(1, 2))
#[40,38,42,39, ...]
normed_air_means = air_means / flat_means.mean() # 40

for i in range(images.data.shape[0]):
    images.data[i] /= air_means[i]

# again if you looked at the means here, they'd all be equal to the mean of the region in the flat images
air_means = images.data[:, air_top:air_bottom, air_left:air_right].mean(axis=1,2)
# [40,40,40,40, ...]   

The compute function for calculating means will need to take a list of arrays, the images and a 1D array to put the means into. It will look something like:
arrays[1][i] = arrays[0][i, air_top:air_bottom, air_left:air_right].mean()

[samtygier-stfc]

Passing the tests now.

A few small tweaks. Also once everything is working the debug print calls can be removed.

[samtygier-stfc]

We can do this a single numpy call:
air_means = images.data[:, region_of_interest.top:region_of_interest.bottom, region_of_interest.left:region_of_interest.right].mean(axis=(1, 2))

Then can remove _calc.mean()

[samtygier-stfc]

Could remove the dict, and just pass back normed_air_means

[samtygier-stfc]

operations_tests.py passes and code looks good. Thanks