The effect of bant selection to success of artificial neural network in hyperspectral classification

In recent years, classification with hyperspectral images is becoming very popular. Development of camera technology is increasing number of researchers who work in this area. Thanks to hyperspectral imaging technology, specific spectral signatures of objects can be handled. Especially, vegetation classification is possible by using the spectral information of hyperspectral images. However due to the change of spectral signatures of vegetations by time and existence of vegetation which exhibit similar spectral properties, finding a generic classification method that can work time independent is difficult. In this study, a neural network-based method which uses bant selected training data is proposed in order to classify corn and cotton vegetation in the land cover. In this study, the effect of band selection for training data set on the success of neural network method is investigated. Results showed that training network with band-selected data is increasing the classification success and makes classification process time-independent.