Closed martinschorb closed 1 year ago
Hi @trautmane
thanks for the quick fix.
Do you have a reasonable set of parameters for CrossCorrelationPointMatchClient
that you could share? I am playing around with some data and can only get it to work with a rather limited fraction of the slices.
Thanks, Martin
Hi @martinschorb ,
We don't use this very often.
Here are the parameters we used a couple of years ago for a FIBSEM stack that had 5 tiles per z-layer arranged all in one row (hence the clipWidth 250):
--baseDataUrl ${BASE_DATA_URL} --owner ${RENDER_OWNER} --collection ${MATCH_COLLECTION}
--renderScale 1.0 --renderWithFilter false --renderWithoutMask false
--clipWidth 250
--ccFullScaleSampleSize 250 --ccFullScaleStepSize 5
--ccMinResultThreshold 0.5 --ccCheckPeaks 50 --ccSubpixelAccuracy true
--matchRod 0.92 --matchModelType TRANSLATION --matchIterations 1000
--matchMaxEpsilonFullScale 2.0 --matchMinInlierRatio 0.0 --matchMinNumInliers 20
--matchMaxTrust 3.0 --matchFilter SINGLE_SET
I think the core code comes from the ImageJ pairwise stitching code, so maybe that might help you understand options.
Hi,
I am trying to see if some of our matches can be derived using cc.
I now run into the following issue:
When I provide
matchMaxEpsilonFullScale
, it runs. I don't know if all client script calls using that parameter require it, but it seems like for CC it should be marked as required.