[pdal-commits] [PDAL/PDAL] ec3a9a: Initial commit of Lloyd's algorithm for k-means cl...

[chambbj]

  Branch: refs/heads/master
  Home:   https://github.com/PDAL/PDAL
  Commit: ec3a9ab58a1fce1bd7a92a3e3910e5a26376a353
      https://github.com/PDAL/PDAL/commit/ec3a9ab58a1fce1bd7a92a3e3910e5a26376a353
  Author: chambbj <brad.chambers at gmail.com>
  Date:   2020-05-27 (Wed, 27 May 2020)

  Changed paths:
    A doc/stages/filters.lloydkmeans.rst
    M filters/FarthestPointSamplingFilter.cpp
    A filters/LloydKMeansFilter.cpp
    A filters/LloydKMeansFilter.hpp
    M filters/private/Segmentation.cpp
    M filters/private/Segmentation.hpp
    M pdal/KDIndex.cpp
    M pdal/KDIndex.hpp
    M pdal/private/KDImpl.hpp

  Log Message:
  -----------
  Initial commit of Lloyd's algorithm for k-means clustering (#3075)

* Initial commit of Lloyd's algorithm for k-means clustering

Default cluster dimensions are XYZ, but user can provide a comma
separated list of alternate dimensions to use.

Initial cluster centers are seeded using farthest point sampling, so
this sampling method has been moved from
filters/FarthestPointSampling.cpp to private/Segmentation.cpp.

Any conditioning or normalization of the dimensions will need to take
place prior to clustering.

KDFlexIndex has been extended slightly to support nearest neighbor
searches on a flexible list of dimensions.

* Recent conda update may have broken things, try pinning libssh2 just to
see

* Remove libssh2 pin

* addressing PR comments