CLASSIFICATION OF NEURAL NETWORKS BASED ON TYPES OF LEARNING
A. Supervised The supervised classification methods are based on user-defined classes and corresponding representative sample sets. The sample sets are specified by training raster data sets, which must be created, prior to entering the process. In supervised classification, spectral signatures are developed from specified locations in the image. These specified locations are given the generic name training sites and are defined by the user. The training data consists of pairs of input objects and desired output. Supervised classification requires the analyst to select training areas where he/she knows what is on the ground and then digitize a polygon within that area. The computer then creates mean spectral signature. Thus, in a supervised classification we are first identifying the information classes which are then used to determine the spectral classes which represent them. Similarly all the pixels are analyzed by the analyst and corresponding spectral signature are created. The Result is Information--in this case a Land Cover map. Common Supervised classifiers:
a. Parallelepiped
b. Minimum distance to mean Maximum likelihood.