D. Wujanz

Determination of Intensity-Based Stochastic Models for Terrestrial Laser Scanners Utilising 3D-Point Clouds

Recent advances in stochastic modelling of reflectorless rangefinders revealed an inherent relationship among raw intensity values and the corresponding precision of observed distances. In order to derive the stochastic properties of a terrestrial laser scanner’s (TLS) rangefinder, distances have to be observed repeatedly. For this, the TLS of interest has to be operated in the so-called 1D-mode—a functionality which is offered only by a few manufacturers due to laser safety regulations. The article at hand proposes two methodologies to compute intensity-based stochastic models based on capturing geometric primitives in form of planar shapes utilising 3D-point clouds. At first the procedures are applied to a phase-based Zoller + Fröhlich IMAGER 5006h. The generated results are then evaluated by comparing the outcome to the parameters of a stochastic model which has been derived by means of measurements captured in 1D-mode. Another open research question is if intensity-based stochastic models are applicable for other rangefinder types. Therefore, one of the suggested procedures is applied to a Riegl VZ-400i impulse scanner, as well as a Leica ScanStation P40 TLS that deploys a hybrid rangefinder technology. The generated results successfully demonstrate alternative methods for the computation of intensity-based stochastic models as well as their transferability to other rangefinder technologies.

Identification of stable areas in unreferenced laser scans for automated geomorphometric monitoring

Current research questions in the field of geomorphology focus on the impact of climate change on several processes causing subsequently natural hazards. Geodetic deformation measurements are a suitable tool to document such 10 geomorphic mechanisms e.g. under the use ofby capturing a region of interest with terrestrial laser scanners that capture a region of interest in a quasi-laminar fashion which results in a so-called 3D point cloud. In this context the centralThe main problem is linked toin deformation monitoring is the transformation of 3D point clouds captured at different points in time (epochs) into a commonstable reference coordinate system. To date, this step has been mostly carried out by usage of artificial targets and/or controlpoints with known coordinates, so called reference points. Several drawbacks are related to 15 this strategy such as the enormous effort to distribute the targets in object spacewithin the area of interest, the required survey by additional geodetic sensors such as total stations or GNSS-receivers as well as the limited extent within the region of interest. In this contribution a surface-based registration methodology is presented, termed the iterative closest proximity algorithm (ICProx), that solely uses pointspoint cloud data as input, similar to the iterative closest point-algorithm (ICP), and hence does not require any artificial targets or extracted geometric primitives, such as planes. The aim of this study was to 20 automatically determineclassify deformations that occurred at a rock glacier, an ice glacier as well as in a rock fall area. For every example two epochs were processed while the ICProx-algorithm’s classification accuracy is 70% on average in comparison to reference data.