feat: bounding box estimation for RealTransformed RealIntervals #37
Conversation
bogovicj
commented
Apr 12, 2022
bogovicj
commented
- add a test
|
This commit provides the functionality discussed in #6 Needs a new release of imglib2 including imglib/imglib2@33c4661 |
* fills spacing argument if few than numDims provided
|
This is waiting on a new imglib2 core release as it depends on imglib/imglib2@33c4661 |
|
@bogovicj I did not look in depth at the code, so forgive me if this is already taken care of... It seems to me, that there can be unfavourable cases, when for example the sampled interval would be 19x19, and the spacing would be 10. Then the max point that would be checked is (10, 10), when we would prefer to also check the actual max corner at (19, 19). I think this is what happens with the current A variant of this is that something similar could happen due to rounding errors with real coordinates, even if the spacing is chosen well-intentioned. |
* fills spacing argument if few than numDims provided
ctrueden
force-pushed
the
realTransformBbox
branch
from
d920a15 to
9a3543a
Compare
|
@bogovicj I rebased this over the latest master branch. It does not compile right now though due to API skew. |
…ransform into realTransformBbox # Conflicts: # pom.xml # src/main/java/net/imglib2/util/RealIntervals.java # src/test/java/net/imglib2/util/RealIntervalsTests.java
axtimwalde
force-pushed
the
realTransformBbox
branch
from
19fe93a to
20c8765
Compare
with extensible interface for sampling method, override trivial cases in Translations, scales, and affine transformations
axtimwalde
force-pushed
the
realTransformBbox
branch
from
20c8765 to
abfad42
Compare
