Spatial-spectral operator theoretic methods for hyperspectral image classification (Q338713): Difference between revisions

The authors analyze data integration and fusion techniques for taking advantage of both spatial and spectral information in hyperspectral imagery. The techniques are based on outcomes of Laplace-type operators on data dependent graphs. The best results are observed when spatial and spectral information is used at all stages: in constructing the data adjacent matrix, and in building the new joint spatial-spectral Laplace operator on such a modified graph.