Abstract: New generation Earth observation satellites have been recently deployed in constellations to increase very significantly the spatial, spectral/radar and temporal resolutions of the digital images acquired over the Earth. The large amount of data collected now raise new challenges for the retrieval of information in these remote sensing image time series. The challenges relate, among others, to handling spectro-angular multimodality, processing long time series of increasingly dense images as well as statistical learning of the evolutions characterizing different dynamic phenomena.  The session will feature a wide range of theory and applications involving both unsupervised and supervised learning frameworks dedicated to Earth observation image time series analysis and satellite computer vision.

Abdourrahmane M. ATTO
Université Savoie Mont Blanc
France