English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In hyperspectral image (HSI) classification, spatial context has demonstratedits significance in achieving promising performance. However, conventionalspatial context-based methods simply assume that spatially neighboring pixelsshould correspond to the same land-cover class, so they often fail to correctlydiscover the contextual relations among pixels in complex situations, and thusleading to imperfect classification results on some irregular or inhomogeneousregions such as class boundaries. To address this deficiency, we develop a newHSI classification method based on the recently proposed Graph ConvolutionalNetwork (GCN), as it can flexibly encode the relations among arbitrarilystructured non-Euclidean data. Different from traditional GCN, there are twonovel strategies adopted by our method to further exploit the contextualrelations for accurate HSI classification. First, since the receptive field oftraditional GCN is often limited to fairly small neighborhood, we proposed tocapture long range contextual relations in HSI by performing successive graphconvolutions on a learned region-induced graph which is transformed from theoriginal 2D image grids. Second, we refine the graph edge weight and theconnective relationships among image regions by learning the improved adjacencymatrix and the 'edge filter', so that the graph can be gradually refined toadapt to the representations generated by each graph convolutional layer. Suchupdated graph will in turn result in accurate region representations, and viceversa. The experiments carried out on three real-world benchmark datasetsdemonstrate that the proposed method yields significant improvement in theclassification performance when compared with some state-of-the-art approaches.

Hyperspectral Image Classification With Context-Aware Dynamic Graph Convolutional Network