Feature Extraction for Hyperspectral Imagery: The Evolution From Shallow to Deep: Overview and Toolbox

Behnood Rasti (1) , Danfeng Hong (2) , Renlong Hang (3) , Pedram Ghamisi (1) , Xudong Kang (4) , Jocelyn Chanussot (5, 6) , Jon Atli Benediktsson (6)

1 HZDR - Helmholtz-Zentrum Dresden-Rossendorf
2 DLR - German Aerospace Center
3 NUIST - Nanjing University of Information Science and Technology
4 HNU - Hunan University [Changsha]
5 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
6 University of Iceland [Reykjavik]

Hyperspectral images (HSIs) provide detailed spectral information through hundreds of (narrow) spectral channels (also known as dimensionality or bands), which can be used to accurately classify diverse materials of interest. The increased dimensionality of such data makes it possible to significantly improve data information content but provides a challenge to conventional techniques (the so-called curse of dimensionality) for accurate analysis of HSIs.

Mots clés: Hyperspectral imaging; Feature extraction; Machine learning; Data mining; Training data

Domaines: Traitement du signal et de l'image [eess.SP]

Behnood Rasti, Danfeng Hong, Renlong Hang, Pedram Ghamisi, Xudong Kang, et al.. Feature Extraction for Hyperspectral Imagery: The Evolution From Shallow to Deep: Overview and Toolbox. IEEE geoscience and remote sensing magazine, 2020, 8 (4), pp.60-88. ⟨10.1109/MGRS.2020.2979764⟩. ⟨hal-03142175⟩