Classification of building attributes in dense urban areas using ALOS-2 data and airborne LiDAR data

In this study, we propose the integration of airborne LiDAR and satellite SAR data for building extraction and classification in four steps. First, we generated a digital surface model (DSM) from airborne LiDAR data. Second, the DSM was registered with a normalized radar cross-section (NRCS) image calculated from the SAR data. Third, buildings were extracted from the DSM, and finally, the buildings were classified into several clusters using NRCS values in the DSM. In our experiment, we selected a dense urban area in Tokyo as our study area. Then, we prepared ALOS-2 PALSAR-2 data and a DSM generated from an airborne LiDAR data. In the building extraction step, we extracted 1778 building roof segments from the DSM. In the classification step, we classified NRCS values of ascending and descending orbit data into several clusters based on ISODATA clustering to estimate building attributes. We conducted an experiment to validate our approach and clarified that a combination of airborne LiDAR and satellite SAR data could extract and classify buildings in a dense urban area.