CLASSIFICATION OF MULTISPECTRAL DATA

This paper deals with the classification of objects into a limited number of classes. Objects are characterised by n-features, e.g. n-dimensional vector are used to describe them. The paper focuses on the Bayes classifier based on the probability principle, with a fixed number of features during the classification process. Bayes classifier, that is which uses the criterion of the minimum error, was applied to the set of the multispectral data. They represented real images of the Earth \u27s surface obtained from remote Earth sensing. This paper describes the experiences and resuIts obtained during the classification of extensive sets of this multispectral data and an analysis of the influence of dispersions and the mean values of the features of the classification results