Texture and Intensity Based Land Cover Classification in Germany from Multi-Orbit Multi-Temporal Sentinel-L Images

摘要

Land cover information is vital for ecosystem management, to find biodiversity indicators and for sustainable development. The launch of the Sentinel-l satellites provide large amounts of Synthetic Aperture Radar (SAR) data that can be used for the extraction and classification of land cover. This study presents a preliminary method for land cover classification using SAR amplitude and textural features and by combining multi-temporal images from ascending and descending orbits. The texture parameters contrast, entropy, homogeneity and variance were investigated. The rules of the SAR-LC classifier, designed and implemented at the Federal Agency for Cartography and Geodesy in Frankfurt, were optimised to include textual information for processing the multi-temporal SAR images. The results for land cover classification of images from 2017 for the area around Berlin in Germany are reported along with their classification efficiencies.