Improvement in measurement accuracy for hybrid scanner

Abstract

The capability to provide dense three-dimensional (3D) data (point clouds) at high speed and at high accuracy has made terrestrial laser scanners (TLS) widely used for many purposes especially for documentation, management and analysis. However, similar to other 3D sensors, proper understanding regarding the error sources is necessary to ensure high quality data. A procedure known as calibration is employed to evaluate these errors. This process is crucial for TLS in order to make it suitable for accurate 3D applications (e.g. industrial measurement, reverse engineering and monitoring). Two calibration procedures available for TLS: 1) component, and 2) system calibration. The requirements of special laboratories and tools which are not affordable by most TLS users have become principle drawback for component calibration. In contrast, system calibration only requires a room with appropriate targets. By employing optimal network configuration, this study has performed system calibration through self-calibration for Leica ScanStation C10 scanner. A laboratory with dimensions of 15.5 m × 9 m × 3 m and 138 well-distributed planar targets were used to derive four calibration parameters. Statistical analysis (e.g. t-test) has shown that only two calculated parameters, the constant rangefinder offset error (0.7 mm) and the vertical circle index error (-45.4") were significant for the calibrated scanner. Photogrammetric technique was utilised to calibrate the 3D test points at the calibration field. By using the test points, the residual pattern of raw data and self-calibration results were plotted into the graph to visually demonstrate the improvement in accuracy for Leica ScanStation C10 scanner.