Support nD images and labels in label2rgb

[Abhinavmishra8960]

#5501

As label2rgb do not support dimension greater than 2. Added Correct error condition and message.

Changed code image

Please do mention changes required.

For reviewers

• Check that the PR title is short, concise, and will make sense 1 year
  later.
• Check that new functions are imported in corresponding __init__.py.
• Check that new features, API changes, and deprecations are mentioned in
  doc/release/release_dev.rst.

[pep8speaks]

Hello @Abhinavmishra8960! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-09-10 17:10:49 UTC

[grlee77]

Thanks @Abhinavmishra8960. What you have here looks okay, but we will be better off making this function support nD. Please see my suggestion on how we can do this with very little work!

After applying that change this would need:

1.) Update the PR title to something like: Add nD support to label2rgb
2.) Add a new test case to skimage/color/tests/test_colorlabel.py with 3D (maybe 1D as well) input.
3.) Update the Parameters section of the docstring in this function to remove reference to "2D".

[grlee77]

One of my prior suggestions caused failures when image was grayscale (:frowning_face:). The new suggestions here fix that and expand the test cases.

[grlee77]

I had to push one fix to the suggested test case. When I took the 1D slice, it was originally only passing through one of the two labels, which resulted in a mismatch in the label color. I updated the label map used in the tests to have both label values intersect the 1D slice.

Assuming the tests pass now, this is ready to go.

[rfezzani]

Seeing this change, I wonder if shouldn't add the support for the channel_axis argument...

[grlee77]

This function does have channel_axis already! I think the question is how to best indicate that labels is "nD spatial" while image is either "nD spatial" or "nD spatial with channels"

[grlee77]

I agree that the docstring as currently written implies that channels are in the last position if present, which may be misleading

[rfezzani]

I suggest to remove the shapes from the label and image variables type and specify in the description that image should have the same shape as label with an optional additional RGB channel axis...

[grlee77]

a pushed a commit with the suggested update. please review

[alexdesiqueira]

Thank you @Abhinavmishra8960 @grlee77!