Interface Software

Point Cloud Registration/Transformation Utility

A free (open source) software utility has been developed at University of Texas at Dallas for registration/alignment of terrestrial LiDAR scans. The software package can apply 6-parameter affine transformation to 3D cartesian data points, as well as derive the correct transformation parameters to apply in order to convert points from one coordinate system to the next -- as, for example, when aligning scans taken from different scan positions or in georeferencing scan data. This is made possible on the basis of common control points whose coordinates are known in both systems (for example, reflectors scanned from both scan positions). A minimum of three points identified in common is required, with at least five recommended.

A common workflow for use of the software might be as follows.

  • Obtain or prepare data files
    • LiDAR [X Y Z] scan data from multiple scan positions
    • Tie-point files indicating locations of control points relative to scan data (with GPS, e.g. UTM coordinates if georeferencing)
  • Plan alignment sequence (in simplest case, choose a single destination coordinate system and align all others to it directly; for wider surveys, alignments may be done transitively)
  • Perform an alignment
    • Compose special tiepoint file showing correspondences between "image" (source) and "reference" (destination) control points
    • Run MATLAB script to derive transformation parameters which, when applied to any point in "image" coordinates, will produce correct location in "reference" coordinates
    • Apply resulting affine transformation (translation and rotation) to all data points presently in "image" coordinates; save the output data and note these points are in "reference" coordinates
    • Confirm, for example by loading aggregate data in a free point cloud viewer, that all data now in "reference" coordinates (original + transformed data) meshes correctly and agrees(analysis of error residuals is also provided for every transformation derivation)
  • Continue alignments until all data is aligned or georeferenced.

Screenshot of software utility

This software package for alignment (including georeferencing) is open source, written in scripting language (MATLAB) and authored by Lionel White. The iterative descent, least squares algorithm for solving the transformation parameters has been documented by the author. Additional tools for the Windows environment may be provided. For more information, please contact Mr. White and/or Jarvis Cline (jrc032000 AT utdallas DOT edu).

 

Resources Home