Structure-from-Motion and Wavelet Decomposition for outcrop analysis
Résumé
The rise of numerical methods in geological sciences, including volcanic surface and subsurface analysis offers the 21st century the opportunity to work on large datasets and refine the level of details collected. At the same time, outcrops are still the base of numerous near-surface analysis, especially on active volcanoes and only few progress has been made around the data-collection method, if not using expensive tools such as terrestrial laser scanners. In the present contribution, the authors have investigated the potential of the photogrammetric method SfM-MFS (Structure from Motion and Multiple-View Stereophotogrammetry ), in order to improve the manual process of outcrop analysis, and they have also investigated the use of wavelets and spatial analysis techniques - usually used in GIS analysis - applied to a vertical surface. Results have proven that SfM-MVS can record precise 3D data from an outcrop. The method also allows the construction of orthophotographs of the outcrop, which can then be used for further image analysis. The usage of wavelets and other spatial algorithms have shown their ability to extract features and variations in the grain-size from numerical data. This study has also shown the limits of such techniques, as in the present technological state, some information collected in the field can't be measured numerically. Therefore the authors have suggested that photogrammetry based algorithms could help extend the analysis potential from one localized control point - where the researcher can sample, measure and describe an outcrop - to a larger area. Hence information from one sample could be extended to long transects and show large/progressive horizontal variations that can't be easily identified from manual work.
Origine : Fichiers produits par l'(les) auteur(s)