Fast and efficient land-cover classification of multispectralremote sensing data using artificial neural network techniques

摘要

A time and memory efficient methodology for supervised andunsupervised land-cover classification of multispectral remote sensing(MRS) data based on artificial neural network (ANN) techniques ispresented. The proposed methodology first performs a vector quantization(VQ) using the self-organizing maps (SOM) algorithm to compress the MRSdata followed by either efficient clustering and automaticclassification or, when training sets are available, by a forcedreduction of the training set size induced by vector quantizationresulting to a faster training of the supervised ANN algorithms