I am looking for the way to compute the principal curvature directions using PCL.
In class PrincipalCurvaturesEstimation the methodc omputePointPrincipalCurvatures() is used for it. It extracts the eigenvalues and eigenvectors from covariance matrix. But it only sets 1 (max) principal direction.
Should I write my own code to get both directions? And could the covariance matrix be used as the curvature tensor?