Modelling of point clouds representing natural surfaces

• Applicant: Professorin Dr. Corinna Harmening
• Subject Area: Geodesy, Photogrammetry, Remote Sensing, Geoinformatics, Cartography
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 564096276

Scientific context: Climate change has a significant impact on the earth's natural surfaces (e.g. glaciers, slopes, fields, coasts). In order to describe these changes and subsequently implement protective measures, the first priority is to monitor the affected areas. Areal measurement methods such as laser scanning have proven to be particularly suitable for these purposes, as the resulting point clouds describe the acquired surfaces quasi-continuously. Usually, the first step in the subsequent data interpretation consists of a modelling of the point clouds, as this reduces the measurement noise and the data volume. The global modelling of point clouds poses a particular challenge, especially for natural objects that are characterised by many local structures. Objectives: The aim of this project is to develop a generally valid modelling approach for point clouds that allows natural surfaces to be described. The approach to be developed is characterised by the fact that the classical tensor product (TP) B-spline surfaces are made applicable for the description of natural surfaces on the basis of suitable parameterisation strategies. In this way, cumbersome refinement strategies can be avoided. Methods: The approximation quality of B-spline surfaces is primarily determined by the preceding parameterization: The assignment of surface parameters to the observed data points localizes them on the surface to be estimated. For the modelling of point clouds representing artificial objects, parameterization strategies in which these surface parameters are determined on the basis of the point cloud’s boundary curves have proven to be favorable. For natural surfaces, however, which are characterized by many local structures, this approach is not expedient. For this reason, a strategy for local parameterization is going to be developed: The resulting assigned surface parameters store the important information about the local variations of the point cloud, so that TP-B spline surfaces can be used in the next step to model the point clouds. Originality and innovation: The modelling of natural surfaces is currently solved with the help of refinement strategies: Starting with classical TP B-spline surfaces, parameters are added to selected areas in order to improve the modelling of the point cloud in these regions. As a result, the B-spline surfaces lose their compact form, which makes further processing more difficult. With the help of the local parameterization to be developed, the compact form of the TP B-spline surfaces can be retained and a satisfactory approximation of the point cloud can be achieved at the same time. This provides an excellent basis for further analysis steps.

• DFG Programme: Research Grants