[pdal] Eigenvalues for .las Data

[Joseph S Faraguna]

Hi everyone,

I am wondering what matrix is passed into Eigen's SelfAdjointEigenSolver
<https://eigen.tuxfamily.org/dox/classEigen_1_1SelfAdjointEigenSolver.html>
when filters.eigenvalues is called on a .las file. The filters.eigenvalues
documentation states that the decomposition is performed for each point
with knn nearest neighbors, but to my understanding decompositions can only
be performed on square matrices. I was originally imagining a matrix of
points like

[x0 x1 x2 x3 ... xknn]

]y0 y1 y2 y3 ... yknn]

[z0 z1 z2 z3 ... zknn]

but this would definitely not be square. Is it a covariance matrix for all
points, kind of analagous to PCA?

Best,

Joe Faraguna
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/pdal/attachments/20190708/384c66b5/attachment.html>