Multi-spectral image classification using adaptive neural network

Multispectral classification is very important in remote sensing areas. Multi-layer backpropagation is used regularly to classify the data. However, it suffers from slow convergence rate for training. In this paper, we are presenting adaptive methods to train the backpropagation network both. locally and globally. The result is a fast convergence rate without any sacrifice on the part of classification accuracy. In addition, we are evaluating the adaptive version of the radial basis function (RBF) method for its versatility of classifying multispectral data.