point data

classification_point data

All classification will be based upon both supervised and unsupervised machine learning with fuzzy logic, because targets are highly affected by imperfect information due to cultural and natural in a temporal context which might expand several thousand years.

A variant over the classic PCA algorithm for adapting n dimensions:

Select a region of points and computer covariance matrix accordingly to xyz. Compute eigenvalues of matrix providing a spread of variance in three mutually orthogonal directions. If eigenvalues are of same magnitude, they are fairly uniformly dispersed. If one eigenvalue close to zero and the other two are not, data is exhibiting planar distribution.

See also: http://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XL-5/187/2014/isprsarchives-XL-5-187-2014.pdf for semi-global matching of planar surfaces in order to understand planar segmentation and classification.

Leave a Reply

Your email address will not be published. Required fields are marked *