pyxem / kikuchipy / 434

100%

push

travis-ci-com

A framework for computing similarity between 2D gray-tone images (#231)

* feat(SimilarityMetric): ZNCC, NDP and user defined metrics

Framework for calculating similarities between 2D gray-tone images of equal size.

* fix(SimilarityMetric): squeeze dimensions of similarity matrix

* fix(SimiliarityMetric): rechunk dask arrays after type conversion

* refactor(SimilarityMetric)

length of shape -> ndim, and removed unnecessary squeeze call.

* test(SimilarityMetric): Many to many ZNCC

Produce value of 1 with equal pattern and template

* refactor: removed metricscopes including ANY

* wip: tests cover most cases

* test: remove print statement

* refactor: indexation -> indexing

* Reformat docstrings, add Ole to credits, add indexing module to doc

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* refactor: renamed variables and made functions private

* refactor: underscore prefix

* refactor: underscore prefix

* Update docstring table, and more

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Add indexing module to kikuchipy/__init__.py

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Add static dictionary indexing note in changelog

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* fix: change standard shape of similarity matrix

Change data shape from (N,nm) to (nm,N) to correspond better with cdist in scipy and general logic.

* test: update tests to new output standard

* fix: _is_compatible to be working for all scopes

* Clarify parameters in docstrings, some minor syntax changes, repr

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Docstring explanation of _is_compatible()

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Return numpy where only numpy arrays are passed, test repr

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Improve return numpy/dask test

Signed-off-by: Håkon Wiik Ånes <hwaanes@gmail.com>

* Change input params ... (continued)

175 of 175 new or added lines in 4 files covered. (100.0%)

5149 of 5261 relevant lines covered (97.87%)

4.89 hits per line