Object‐based classification for mountainous vegetation physiognomy mapping
Classification de la physionomie paysagère de montagne par classification supervisée orientée « objet »
Résumé
Montane landscape physiognomy consists mainly of natural and semi-natural elements where ecological variability, extreme meteorological conditions and anthropogenic influence make it highly heterogeneous on multi-scale levels. Thus, very high spatial resolution (VHSR) optical satellite images appear indispensable for describing the physiognomic composition of mosaic-like landscape and for facilitating land cover mapping on large areas. This chapter considers this mosaic-like landscape as open environments starting at the subalpine zone. It presents an application of object-based classification for montane landscape physiognomy mapping. Automatic remote sensing detection of montane vegetation physiognomy is based on VHSR images. The chapter considers a pansharpening RCS method implemented in the Orfeo Toolbox (OTB) library. It presents a geographic information system method in order to use a vector layer of samples produced independently from a pre-produced segmentation, that is by manual digitization and photo-interpretation.