Centre dedocumentation
scientifique

Cross-guided pyramid attention-based residual hyperdense network for hyperspectral image pansharpening / Jiahui Qu in IEEE Transactions on geoscience and remote sensing, vol 60 n° 11 (November 2022)  

Public [article]  inIEEE Transactions on geoscience and remote sensing > vol 60 n° 11 (November 2022) . - n° 5543114 Titre : Cross-guided pyramid attention-based residual hyperdense network for hyperspectral image pansharpening Type de document : Article/Communication Auteurs : Jiahui Qu, Auteur ; Tongzhen Zhang, Auteur ; Wenqian Dong, Auteur ; et al., Auteur Année de publication : 2022 Article en page(s) : n° 5543114 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] attention (apprentissage automatique) [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] image hyperspectrale [Termes IGN] image panchromatique [Termes IGN] lissage de données [Termes IGN] pansharpening (fusion d'images) Résumé : (auteur) Hyperspectral (HS) image pansharpening is of great importance in improving the spatial resolution for many commercial platforms and remote sensing tasks. Convolutional neural network (CNN) has recently been applied in pansharpening. However, most existing CNN-based pansharpening models followed an early-fusion/late-fusion strategy, which integrates the low-level/high-level features of panchromatic (PAN) and HS streams at the input-output of the network. It is difficult to learn more complex combinations between PAN and HS streams. This article proposes a novel end-to-end residual hyperdense pansharpening network with a cross-guided pyramid attention (called RHDcgpaNet). The overall architecture of the proposed method is a residual hyperdense network, which extends the definition of dense connections to two-stream pansharpening problem. The proposed RHDcgpaNet allows guidance from the state of the preceding layers to all the layers in- between PAN and HS streams in a feed-forward manner, significantly increasing the learning representation. A cross-guided pyramid attention is designed and embedded to the proposed residual hyperdense network to yield more useful spatial–spectral feature transfer in network. Extensive experiments on widely used datasets demonstrate that the proposed RHDcgpaNet achieves favorable performance in comparison to the state-of-the-art methods. Numéro de notice : A2022-852 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.1109/TGRS.2022.3220079 Date de publication en ligne : 07/11/2022 En ligne : https://doi.org/10.1109/TGRS.2022.3220079 Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=102098 [article]