WIP: add a new segmentation module

[poilvert]

@npinto,

First draft for a new segmentation module. So far I needed the pad2d function from
sthor so I added it to the module, even though in the future we will remove that extra
code and depend on sthor.

I also made a proxy to the Adjusted Rand Index metric from sklearn that seems fast
enough. That metric though, is only usable on segmented images (on needs multiple
labels for each connected component).

I added a quick build script to build a small Cython module that I use in the warping error
code.

Finally here are the steps to make it (hopefully) work :

1. build the Cython extension module by executing build_unique.sh
2. run the demo code warping_error_demo.py

Cheese,

N.

[npinto]

Nice! I learned something here, I missed unravel_index from numpy >= 1.6.0 !

[npinto]

Looks awesome! I think we are missing the following

• clean cython build (see https://github.com/npinto/sthor/blob/develop/sthor/operation/setup.py#L7)
• smoke tests (Grab'ed)
• tests with random and pseudo-random images
• more complete tests based on Fiji (Java implementation).
• demo in examples
• (speed) benchmarks (if needed)
• topology-breaking points (ENH)

Anything else?

[npinto]

image_true = (image_true > 0).astype(...)

[poilvert]

@npinto

see if that fits you. I have some examples also in examples

[npinto]

Up ?

[npinto]

hardcoded path again here...

[npinto]

demo for ha (please discard)

python code