Semantic Segmentation of Metoceanic Processes Using SAR Observations and Deep Learning

Remote Sensing

Romain Husson¹  Charles Peureux¹  Aurélien Colin²  Pierre Tandeo²  Ronan Fablet²  Alexis Mouche³  Nicolas Longépé⁴ 

[1] Collecte Localisation Satellites, F-31520 Brest, France;IMT Atlantique, Lab-STICC, UMR CNRS 6285, F-29238 Brest, France;Laboratoire d’Oceanographie Physique et Spatiale, Ifremer, F-31520 Brest, France;Φ-Lab Explore Office, ESRIN, European Space Agency (ESA), F-00044 Frascati, Italy;

关键词: SAR;    segmentation;    metocean;    deep learning;    supervised learning;    weakly-supervised learning;   

DOI  :  10.3390/rs14040851

来源: DOAJ

【 摘 要 】

Through the Synthetic Aperture Radar (SAR) embarked on the satellites Sentinel-1A and Sentinel-1B of the Copernicus program, a large quantity of observations is routinely acquired over the oceans. A wide range of features from both oceanic (e.g., biological slicks, icebergs, etc.) and meteorologic origin (e.g., rain cells, wind streaks, etc.) are distinguishable on these acquisitions. This paper studies the semantic segmentation of ten metoceanic processes either in the context of a large quantity of image-level groundtruths (i.e., weakly-supervised framework) or of scarce pixel-level groundtruths (i.e., fully-supervised framework). Our main result is that a fully-supervised model outperforms any tested weakly-supervised algorithm. Adding more segmentation examples in the training set would further increase the precision of the predictions. Trained on 20 × 20 km imagettes acquired from the WV acquisition mode of the Sentinel-1 mission, the model is shown to generalize, under some assumptions, to wide-swath SAR data, which further extents its application domain to coastal areas.

【 授权许可】

Unknown