Semi-supervised hyperspectral pixel classification using interactive labeling

A semi-supervised pixel classification scheme for hyperspectral satellite images is presented. The scheme includes a previous band selection step followed by a clustering process to select modes of interest that will be labeled by an expert. Then pixel classification is performed resulting in a segmentation and classification of the fields appearing in the image. Thanks to the previous clustering step the most suitable pixels are automatically selected to build the classifier. This reduces the expert effort required since less pixels need to be labeled. However pixel classification accuracy obtained outperforms the results of a random selection scheme where many more pixels were labeled.