|
Monitoring and maintenance of protective infrastructure assests is a key aspect of common crisis prevention strategies in the alpine area. However, the current gold standard of manual inspection and surveying is labour intensive, demanding, and sometimes dangerous to the inspectors. Moreover, collected data usually is of qualitative rather than quantitative nature and downstream analysis options are relatively limited. In this work, we introduce an easy-to-use, operator-assumption-light approach to extract check dam structures from high density multi-echo aerial laserscans using a combination of 3D point cloud processing techniques and morphological filtering in the 2.5D space. The only required inputs to the anayltical pipeline are a georeferenced point cloud of the scanned area as well as the approximate position of the check dams – an information usually readily available from geographic information systems (GIS). Another strength of our method lies in its ability to cope with locally sparse or missing data as well as dense vegetation overgrowing the check dams. In general, the automatically reconstructed checkdam structures are in good agreement (deviation < 10 mm) with manually cropped data, albeit with some considerable deviations around edges of near-vertical parts of up to several meters as a result of erosive properties of the morphological filters. In addition to reconstruction of check dams and their surrounding terrain, we also estimate the overflow level of the check dams. This allows us to automatically obtain their respective upstream reservoirs from the surrounding terrain using region growing techniques. We see the digital model thus obtained as being well suited to support rapid quantitative assessment of the remaining protective properties of a given check dam and by extension maintenance related decision-making.
|
