ENH: ScaleAnImage Python example and input data update

[tbirdso]

• Adds Python example and baseline test
• Replaces contrived square image input generated at runtime with static cthead1.png input

Input

Output

[thewtex]

Nice! A few Pythonic improvements noted inline.

[tbirdso]

Seeing a difference in baseline comparisons between cxx and Python examples despite attempting to run the exact same set of procedures for image scaling. Cxx output matches baseline (which was generated with the cxx test) and appears visually correct, Python output does not appear to scale the image border as expected.

I've tested both itk.ResampleImageFilter class directly and the itk.resample_image_filter functional interface, with the same results. @dzenanz do you have any thoughts on possible reasons why we might get different output from similar input to itk::ResampleImageFilter?

Input

Cxx output

Python output

[dzenanz]

If float/double difference is not it, I don't know what might be causing it.

[tbirdso]

Closing this stale PR as I haven't been able to pinpoint the issue source and the issue itself seems trivial as it is a difference of one pixel width. Unfortunately there isn't a good way to reference a different baseline image for Python vs cxx tests here.

If this is seen as a symptom of a larger problem (Python vs cxx behavior difference) we can re-submit as an issue against ITK.

[dzenanz]

An open PR is visible, and people can "steal" code from it easier. If we don't have need/time to resolve the issue now, we should keep it open until we do, or until someone else suggests a fix.

[tbirdso]

Is it possible this is related to ITK Issue 2703?

Note that the root of discussion there is ResampleInPlaceFilter (docs) and focuses on image direction discrepancies while this example uses ResampleImageFilter and discrepancies in pixel data in the output image. It does not appear that these two filters inherit from each other.

[tbirdso]

Pinging @brad-t-moore if he is available to get another set of eyes on this issue

[dzenanz]

I don't think this is related to issue 2703.

[tbirdso]

@thewtex Ping for re-review

[thewtex]

Looks great!

Minor style nitpick inline

[tbirdso]

Resolved merge conflicts.