This study proposes a new metric to process dual-polarimetric coherent and incoherent synthetic aperture radar (SAR) data for coastline extraction purposes. The metric, based on the correlation between co- and cross-polarized channels, allows discriminating land from sea in an unsupervised way. Then, simple image processing is adopted to extract continuous coastline from the binary image. Experiments, undertaken on multipolarization C- (RadarSAT-2 and Sentinel-1) and X-band (Cosmo-SkyMed) SAR data collected in South of Italy together with Global Positioning System ground truth, confirm the soundness of the method which is shown to be both effective (a whole SAR scene is processed in seconds) and accurate (the mean error is less than 5 and 7 pixels for RadarSAT-2 and CosmoSkyMed, respectively).
Dual-Polarimetric C- and X-Band SAR Data for Coastline Extraction / Nunziata, Ferdinando; Buono, Andrea; Migliaccio, Maurizio; Benassai, Guido. - In: IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING. - ISSN 1939-1404. - (2016), pp. 1-8. [10.1109/JSTARS.2016.2560342]
Dual-Polarimetric C- and X-Band SAR Data for Coastline Extraction
NUNZIATA, FERDINANDO;MIGLIACCIO, Maurizio;
2016
Abstract
This study proposes a new metric to process dual-polarimetric coherent and incoherent synthetic aperture radar (SAR) data for coastline extraction purposes. The metric, based on the correlation between co- and cross-polarized channels, allows discriminating land from sea in an unsupervised way. Then, simple image processing is adopted to extract continuous coastline from the binary image. Experiments, undertaken on multipolarization C- (RadarSAT-2 and Sentinel-1) and X-band (Cosmo-SkyMed) SAR data collected in South of Italy together with Global Positioning System ground truth, confirm the soundness of the method which is shown to be both effective (a whole SAR scene is processed in seconds) and accurate (the mean error is less than 5 and 7 pixels for RadarSAT-2 and CosmoSkyMed, respectively).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
