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

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