Add examples to scipy.ndimage.filters

[MartinThoma]

No description provided.

[rgommers]

Import style: from scipy import ndimage, misc

[rgommers]

And a minor thing for next commits: please use the prefixes given in http://docs.scipy.org/doc/numpy-dev/dev/gitwash/development_workflow.html#writing-the-commit-message

[rgommers]

Overall +1, nice to have some examples for ndimage.

[MartinThoma]

Thank you for pointing me to the commit prefixes; I adjusted my commit message (and changed the other things you mentioned).

[rgommers]

Code looks good now, but the example is a pretty bad green: https://circle-artifacts.com/gh/scipy/scipy/145/artifacts/0/home/ubuntu/scipy/doc/build/html-scipyorg/generated/scipy.ndimage.prewitt.html#scipy.ndimage.prewitt

How about adding plt.gray() before plt.imshow to show the filtered result in grayscale? Another colormap is fine with me too, but the current default is not great.

[MartinThoma]

@rgommers Thank you for the hint; I've added it: https://circle-artifacts.com/gh/scipy/scipy/146/artifacts/0/home/ubuntu/scipy/doc/build/html-scipyorg/generated/scipy.ndimage.prewitt.html#scipy.ndimage.prewitt

[rgommers]

That looks much better. One last issue - the second figure for gaussian_laplace doesn't show up: https://circle-artifacts.com/gh/scipy/scipy/146/artifacts/0/home/ubuntu/scipy/doc/build/html-scipyorg/generated/scipy.ndimage.gaussian_laplace.html#scipy.ndimage.gaussian_laplace

If you insert a plt.figure() above the last 2 lines it probably will. Then I think you'll get 2 figures below each other. Better would be to have them next to each other - this can be done by creating 2 axes in a figure, like:

fig = plt.figure()
plt.gray()
ax1 = fig.add_subplot(121)
ax1.imshow(...)
ax2 = fig.add_subplot(122)
ax2.imshow(...)

[MartinThoma]

Yes, that looks nicer: https://circle-artifacts.com/gh/scipy/scipy/152/artifacts/0/home/ubuntu/scipy/doc/build/html-scipyorg/generated/scipy.ndimage.gaussian_laplace.html#scipy.ndimage.gaussian_laplace

[rgommers]

Much better. In it goes. Thanks @MartinThoma!