Self-calibration and direct georeferencing in terrestrial laser scanning

University dissertation from Stockholm : KTH

Abstract: An important step in data processing from terrestrial laser scanning (TLS) is georeferencing, i.e. transformation of the scanner data (point clouds) into a real world coordinate system, which is important for their integration with other geospatial data. An efficient approach for this is direct georeferencing, whereby the position and orientation of the scanner can be determined in the field, similarly to the working routine of total stations. Thus the efficiency of the survey can be increased, and the project time and costs reduced. An important factor that affects the results of TLS surveys, especially those with direct georeferencing, is scanner calibration. In the recent years, the method of self-calibration used in photogrammetry has become popular for the recovery of systematic errors in laser scanners.This thesis has two main aims. The first one is to develop an approach for self-calibration of terrestrial laser scanners, which can be made available to users, and apply it to the calibration of a number of pulsed laser scanners in order to get a better insight into the systematic instrumental errors present in these instruments. The second aim is to investigate the possibilities for direct georeferencing in TLS in static applications, with the focus on the use of GPS for this purpose, and to develop a survey system based on the combination of TLS and GPS. An additional aim of the thesis is to make a systematic description of the error sources in TLS surveys, where direct georeferencing is employed. A good understanding of these error sources is necessary to secure the data accuracy. We subdivide these errors into four groups: instrumental, object-related, environmental and georeferencing.We have developed a unified approach for self-calibration of terrestrial laser scanners, where one can introduce stochastic information about all the estimated parameters, which helps in reducing their correlations. In part, it is possible to use direct georeferencing to determine the exterior orientation parameters of the scanner. We applied this approach to the self-calibration of the pulsed scanners Callidus CP 3200, Leica HDS 3000 and Leica Scan Station. The initial assumption was that the scanner systematic instrumental errors, or calibration parameters, were similar to those in a total station. However, other errors not explained by the “a priori” total station error model can be present in the scanners. We revealed two such errors – the scale errors in the vertical angles and horizontal directions in the scanners Callidus CP 3200 and Leica HDS 3000, respectively. Most systematic errors were estimated with relatively high precision and low correlations with other system parameters.We have developed a prototype combined survey system, which allows the user to use GPS for direct georeferencing of the scanner parallel to the scanning. In the current implementation, the system consists of the scanning system Leica Scan Station 2, 2 GPS receivers and antennas from Leica and a number of necessary accessories. The scanner position can be determined from RTK (or possibly Network-RTK) measurements with the accuracy of better than 1 cm, both in plane and height. The position of the backsight target can be determined from post-processing of static GPS measurements with similar accuracy. In order to estimate the accuracy of the combined system and its efficiency in a typical TLS survey, we carried out several test measurements. The results have shown that it is possible to achieve the coordinate accuracy of better than 1 cm at the object distance of up to 50 m. This is comparable to the accuracy of conventional direct georeferencing, i.e. when the scanner is centred over a known point. The time expenses for the test survey of a building located at KTH campus, starting from the planning and finishing with the georeferenced point cloud, were about 1.5 workdays. The time expenses could be reduced further if the system was installed on a moving platform during the fieldwork. Hence, the combined system can be successfully used for the surveys of built environments, e.g. engineering constructions and historical monuments, which can be carried out fast and with high accuracy.