Supervised classification is the process of automatically grouping data into a set of classes by setting up prototypes using a priori knowledge obtained through training. It often involves selection of training data that can represent homogeneous examples of each class. There are a number of ways to collect training data, including on-site fieldwork, manual interpolation, automatic seeding, and personal experience. The parameters (typically, mean vector and covariance matrix in multispectral images) of the known training data are calculated for each class and then used for inputs to the classification algorithm for evaluating and assigning the rest of the data to classes.

Various classification algorithms (or classifiers) may be used in supervised classification to assign a data value to a class. These algorithms can be categorized into ...