Repository navigation
Handle multi geometric feature computation - #140
Conversation
| virtual void reset(); | ||
|
|
||
| //! PCA results (populated by computePCA, cleared by reset) | ||
| PCAResults pca; |
There was a problem hiding this comment.
Should be named m_pca, and it should be declared protected at the end of the class definition with the other members
There was a problem hiding this comment.
But you can probably add a getter if necessary.
There was a problem hiding this comment.
I think will have to create local copies of eigen values and eigen vectors anyway, in order to sort them with SortEigenValuesAndVectors for the feature computation.
Unless we cache by default sorted values to avoid any copies. computeCurvature and computeMomentOrder1 don't seem to be affected by this, however I think that it's cleaner to cache the original unsorted.
| { | ||
| m_structuresValidity = FLAG_DEPRECATED; | ||
| pca = PCAResults(); | ||
| m_pca = PCAResults(); |
|
I just realized the way you manage the PCA member is not consistent with the others (i.e. with a flag, and getters that return a pointer if the auto-computation is successful). I'm going to push a commit to make things consistent. |
I see, yes. Thanks. |
Handling the computation of multiple geometric features at one octree traversal.
Associated with the PR #2489 in CloudCompare