Application of supervised neural network approaches to remotely sensed optical imagery

This paper presents a new method based on the application of modular neural networks to agricultural land use classification and compares the advantages and disadvantages over a single complex neural network approach. Neural networks (NN) have been found to have good generalisation properties and their use is becoming increasingly prevalent in the field of remote sensing. However, there are a number of remote sensing problems where neural networks do not necessarily provide an optimum solution, these include mixed pixel analysis, subclass characterisation and parameter extraction for use in biophysical models. Typically the application of NN techniques to remote sensing involves using one NN to classify a large number of land-cover classes. The authors have previously found this approach to be inefficient and inaccurate, a modular approach is therefore implemented which is more flexible. This paper applies these techniques to optical imagery. The area used for this work is a research farm in Bavaria, Germany, which comprises of a highly dynamic terrain with small field units. High resolution land-use maps and yield data have been produced for the research farm, using GPS equipment attached to crop harvesters. These maps enable accurate selection of test classes and are used to validate the results produced by the various techniques