Processing Landsat TM data using complex-valued NRBF neural network

This paper describes a novel classification technique - a complex-valued normalized radial basis function (NRBF) neural network classifier. Complex-valued weights are used in the supervised learning part of NRBF neural networks. Different from the original NRBF neural network, another activation function for the output was added in NRBF neural network. This new neural network model improves the classification ability of NRBF neural networks regardless of the learning method in the unsupervised part. This classifier was tested with satellite multi-spectral image data. Classification results show that this new neural network model is more accurate and powerful than the conventional NRBF model and can solve classification problems more efficiently.