Implemented the Fourier Normalizing Radial Gradient Filter

[vatch123]

Description

It implements Fourier normalizing-radial-graded filter (FNRGF) as stated here.
The changes made are -

• Imported sunpy.coordinates.frames
• Changes in the normalizing_radial_gradient_function
  • Changed the default intensity_summary method from numpy.mean to numpy.nanmean
  • Changed the keyword arguments from None to empty dictionary because it raised an error at runtime stating the types did not match
• Implemented fourier_normalizing_radial_gradient function.
  • The function seems to work fine for different values of order and ratio_mix. Though I am unable to guess what values of parameters are best suited.
  • The default values of few parameters except for the ratio_mix is set by taking inspiration from the doctoral thesis.
  • Some initial tests are added, which needs further improvement.

TODO:

• Review of PR
• ~100% test line coverage
• Code formatted/linted.
• Example and documentation is detailed.

Fixes #7

[pep8speaks]

Hello @vatch123! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file examples/radial_gradient_filters.py:

Line 6:101: E501 line too long (105 > 100 characters)

• In the file sunkit_image/radial.py:

Line 315:101: E501 line too long (104 > 100 characters)
Line 316:101: E501 line too long (105 > 100 characters)
Line 319:101: E501 line too long (105 > 100 characters)
Line 324:101: E501 line too long (106 > 100 characters)
Line 325:101: E501 line too long (106 > 100 characters)
Line 333:101: E501 line too long (113 > 100 characters)
Line 336:101: E501 line too long (113 > 100 characters)
Line 344:101: E501 line too long (105 > 100 characters)
Line 345:101: E501 line too long (106 > 100 characters)
Line 346:101: E501 line too long (113 > 100 characters)
Line 358:50: E203 whitespace before ':'
Line 383:101: E501 line too long (106 > 100 characters)
Line 384:101: E501 line too long (101 > 100 characters)
Line 385:101: E501 line too long (101 > 100 characters)
Line 386:101: E501 line too long (104 > 100 characters)
Line 387:101: E501 line too long (105 > 100 characters)
Line 388:101: E501 line too long (105 > 100 characters)
Line 401:101: E501 line too long (106 > 100 characters)
Line 422:101: E501 line too long (104 > 100 characters)

Comment last updated at 2019-06-10 14:50:17 UTC

[ghost]

Thanks for the pull request @vatch123! Everything looks great!

[nabobalis]

Could you run black, isort and docformater on your PR?

[vatch123]

Could you run black, isort and docformater on your PR?

Yes, it will be my final commit after everything is done.

[nabobalis]

Suggested change

	This function sets the attenuation coefficients in the one of the following two manners-
	This function sets the attenuation coefficients in the one of the following two manners:

[nabobalis]

Can you try to change the plotting values to prevent the image from being so washed out?

I do wonder if something is wrong with this version of the algorithm.

[vatch123]

Okay, which values?
But I don't think the algorithm does anything other than what was intended

[nabobalis]

Not sure, some of the plotting values?

[nabobalis]

Codecov seems to suggest that these two lines are not triggered by the tests:

if cutoff > (order + 1):
    raise ValueError("cutoff cannot be greater than order + 1")

https://codecov.io/gh/sunpy/sunkit-image/compare/118447604c51a28de4d9390511f5d1787f5d8a1d...72b95663cdb182a34f9c214b9c5f3f73618bd7b9/diff#D10-355

and

if order < 1:
     raise ValueError("Minimum value of order is 1")

https://codecov.io/gh/sunpy/sunkit-image/compare/118447604c51a28de4d9390511f5d1787f5d8a1d...72b95663cdb182a34f9c214b9c5f3f73618bd7b9/diff#D10-448

Could you write a test to trigger those warnings and check the messages are correct?

[nabobalis]

This is very close now.

[vatch123]

I hope its done now!!

[nabobalis]

Yes I think that addresses everything!

[nabobalis]

Thanks @vatch123