On-site Self-calibration Using Planar Features for Terrestrial Laser Scanners

Terrestrial laser scanners provide a three-dimensional sampled representation (i.e. point cloud) of the surfaces of objects. They have great potential to improve the measurement and representation of remote and widespread objects for applications such as engineering metrology, cultural heritage recording and forestry, among others. Prior to performing measurement tasks such as these, proper error modelling and estimation is essential in order to remove the inherent systematic effects such as range finder offset, collimation axis error, etc.. A rigorous, point-based self-calibration method has been demonstrated to be effective, but it is very labour-intensive since it requires manual measurement of a large number of signalised targets. In this paper, we propose a planar-feature-based “on-site” self-calibration method that can reduce the manual labour needed in the point-based method. After outlining the principles and mathematical models of the proposed method, the subject of model identification is addressed. Tests with simulated datasets reveal that the residual patterns from the plane-based method are markedly different from those of the point-based method. The ramification of this outcome is that systematic error identification, an important process for new instrumentation such as terrestrial laser scanners, is not straightforward. In addition, the tests of the proposed method with real terrestrial laser scanner datasets are presented and analysed.