Reconstructing surfaces from LIDAR data

ARANZ's FastRBF^TM software and surfacing tools have been successfully applied to the otherwise difficult problem of reconstructing smooth and continuous surfaces from LIDAR scans.

LIDAR data is often noisy, irregular, incomplete and redundant, due to:

• limited range resolution
• misregistration between scans taken at different scanner positions
• restricted physical viewing angles for the scanner
• increased susceptibility to occlusion.
• over-lapping scans

The nominal accuracy for a LIDAR scan is 0.006m, but this excludes scan registration errors and other sources of error. Focussing, calibration and surface roughness, as well as movement in the field of view, can lead to additional errors.

Example of a LIDAR reconstruction of a gate using FastRBFTM's error-bar fitter*.

FastRBF's unique error-bar fitter fits the smoothest surface within tolerances (error bounds) specified at each data point.

• 1,665,500 data points
• 283,200 RBF centres
• Height 15m
• Fit tolerance (error bar) 0.005m
• Isosurface resolution 0.05m

Examples

• Error bar fit to statue of liberty LIDAR data
• Spline smoothing fit to a LIDAR scan
• Smoothing and approximation FAQ
• Surface fitting FAQ
• Surfacing comparison of low pass filter, error-bar fit and spline smoothing
• Surfacing applications

^{*Lidar data courtesy of Cyra Technologies}