Supervised classification

In supervised classification, training data is used for classification. This training data is made in such a way that it is representative of the classes or land cover types we want to classify. An unclassified image is classified using the spectral signature of the pixels in the training data or area. There are three main supervised classification algorithms that are used in QGIS: minimum distance, maximum likelihood (ML), and spectral angle mapper (SAM). For minimum distance, a pixel is assigned to a class that has a lower Euclidean distance to mean vector of a class than all other classes. In ML, each pixel is assigned to the class that has the highest probability. The SAM algorithm works by computing the angle between the mean vector of the class and the unclassified raster data, and the class for which the angle is the smallest is assigned to be...