DOC: adjusted to fit more general case n != m#83
DOC: adjusted to fit more general case n != m#83thewtex merged 2 commits intoInsightSoftwareConsortium:masterfrom
Conversation
|
Also, the output is a scalar image no matter how many channels the input has (which is the reason for the filter and the "magnitude" in its name), see e.g. the example WatershedSegmentation1.cxx |
|
How to see in the ci/circleci page ("Details" links) what caused the failure? I only see that |
|
@romangrothausmann thanks for contributing this fix! We have now officially migrated to GitHub! Could you please rebase this patch with something like: ? |
dzenanz
left a comment
There was a problem hiding this comment.
Thanks for the contribution. As this is a purely documentation change, there should be no test CI failures. As Matt said, rebasing on current master and re-pushing will re-trigger all the checks, which should now all be OK.
adjusted documentation to fit the more general case for images of dimension N with M input channels, especially n != m.
1eaf416 to
bffc8f8
Compare
thewtex
left a comment
There was a problem hiding this comment.
Thanks for the branch update @romangrothausmann , and the documentation correction 👍 Looks good.
The Windows failure is a false positive.
|
Many thanks @thewtex and @dzenanz for taking a look and approving, and also for making contribution to ITK via GH possible. I find it much easier this way, especially the barrier to propose a contribute is much lower with GH PRs and the GH front-end allows many more ways to investigating the project and its code. |
|
Ease of contribution was a major point of the move to GitHub! |
COMP: Update CI for ITK 5.4.4
To my understanding eigenvalues and PCA are the special case for a m×n matrix with n = m, i.e. a square matrix. However, the first order derivative (Jacobian) can be a non-square matrix, i.e. m ≠ n, in case the image dimension does not equal the number of vector components (channels).
The PR is meant as a discussion initiation, as I am not an expert in this field and stumbled upon this while trying to understand the math behind this filter.