You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When using rowan.mapping.icp, the algorithm returns a matrix R and a translation t that map the points in set X to set Y. However, if the points in X and Y are permuted (re-ordered), then it's not enough to rotate and translate the points - they must also be permuted.
The permutation of indices that minimizes the mapping is used internally in the Iterative Closest Point algorithm, and it would be helpful to return those indices alongside the rotation matrix R and translation t.
The text was updated successfully, but these errors were encountered:
As we discussed offline, the internal indices are just based on the closest points found when comparing the two configurations. The main value in returning the indices is simply to provide a way to more effectively measure how good the match was. I'm open to returning the indices based on an optional argument (i.e. adding return_indexes=False to the signature) if there's a desire to have that available.
@bdice what are your thoughts? I'm not really convinced that this is necessary, but since it's essentially free I would be fine with a PR to optionally return the indexes.
When using
rowan.mapping.icp
, the algorithm returns a matrixR
and a translationt
that map the points in setX
to setY
. However, if the points inX
andY
are permuted (re-ordered), then it's not enough to rotate and translate the points - they must also be permuted.The permutation of indices that minimizes the mapping is used internally in the Iterative Closest Point algorithm, and it would be helpful to return those indices alongside the rotation matrix
R
and translationt
.The text was updated successfully, but these errors were encountered: