Extended LVQ neural network approach to land cover mapping

A competitive neural network (NN) learned by the OLVQ1 algorithm has the potential to produce images for land cover mapping. We propose an extended OLVQ1 that adds learning ratios to the original algorithm. Applying it to SPOT[center dot]XS data, we show that higher mapping accuracies can be obtained compared to those of conventional methods. Index Terms - Land cover mapping, learning vector quantization (LVQ), SPOT[center dot]XS data.