Deprecated
This project is
deprecated. It's content has been moved to
dask-image, which users are
encouraged to switch to and redirect all inquiries towards. This organization
remains only for historic value.
Intro
This organization includes a collection of libraries that perform basic image
processing routines on
Dask arrays. Using
Dask arrays makes it easier define computations that work out-of-core. Also
it makes it easier to parallelize Python code without having to leave
Python.
Purpose
The main objective of this organization is to provide a reasonable subset of
functions from SciPy's
ndimage
package implemented so as to work with Dask arrays. The implementations here
do this by either wrapping the existing SciPy ndimage functions or by
reimplementing them with equivalent operations on Dask arrays. In either
case, their results are tested against equivalent functions from SciPy
ndimage to ensure they are being faithfully reproduced.
Caveats
While the initial goal is to provide Dask array implementations of all of
these functions, libraries may be released before having implemented all
functions, or all options to all functions, or without applying some
constraints to the functions. This is done because this is seen as an
incremental work that is driven by research needs. That said, the goal is to
have feature complete implementations of these functions in the end.
Libraries
Below is a list of the libraries that implement the different sets of
functions from SciPy ndimage with Dask arrays. Links to their repos are
provided so that more info about them can be found.