NEURAL NETWORK CLASSIFICATION OF DIFFICULT-TO-DISTINGUISH TYPE OF VEGETATION ON THE BASIS OF HYPERSPECTRAL FEATURES
E. S. Nezhevenko
Institute of Automation and Electrometry, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Keywords: классификация, гиперспектральное изображение, преобразование Гильберта-Хуанга, главные компоненты, нейронные сети, classification, hyperspectral image, Hilbert-Huang transformation, principal components, neural networks
Abstract
It is experimentally demonstrated that the classification of fragments of a hyperspectral images with preliminary transformation of the spectral features of the image into the principal components and with the use of the Hilbert - Huang spectral transformation is fairly effective in the case of vegetation types that are difficult-to-distinguish on the basis of hyperspectra. This classification is compared with traditional methods, where hyperspectral features transformed to the principal components without using spatial information are used. Neural networks RBF are used in all methods at the final stage of the classification.
|
