TomoSAR is widely used for monitoring forests, playing a critical role in climate tracking and biomass measurement. Multiple-baseline acquisitions give TomoSAR the ability to get 3D information on forested areas. However, more acquisitions result in high costs, heavy workloads, and complex signal and image processing techniques. Thus, to demonstrate the possibility of realizing the forest and ground height estimation with less number of baselines. In this study, we utilize TSNN to estimate forest height and underlying topography using multi-polarization and various multi-baseline data options. We conduct experiments with baseline configurations ranging from 5 to 3 to explore TSNN's potential. The experimental results demonstrate that TSNN performs well in height estimation with 5-baseline data, while TSNN trained with 4-baseline and 3-baseline data provides acceptable accuracy.
Deep Learning based PolTomoSAR for Forest Reconstruction
Yang W.;Vitale S.;Aghababaei H.;Ferraioli G.;Schirinzi G.;Pascazio V.
2024-01-01
Abstract
TomoSAR is widely used for monitoring forests, playing a critical role in climate tracking and biomass measurement. Multiple-baseline acquisitions give TomoSAR the ability to get 3D information on forested areas. However, more acquisitions result in high costs, heavy workloads, and complex signal and image processing techniques. Thus, to demonstrate the possibility of realizing the forest and ground height estimation with less number of baselines. In this study, we utilize TSNN to estimate forest height and underlying topography using multi-polarization and various multi-baseline data options. We conduct experiments with baseline configurations ranging from 5 to 3 to explore TSNN's potential. The experimental results demonstrate that TSNN performs well in height estimation with 5-baseline data, while TSNN trained with 4-baseline and 3-baseline data provides acceptable accuracy.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
