11.5 Decision Tree Classifier

The decision tree classifier is an hierarchically based classifier which compares the data with a range of properly selected features. The selection of features is determined from an assessment of the spectral distributions or separability of the classes. There is no generally established procedure. Therefore each decision tree or set of rules should be designed by an expert. When a decision tree provides only two outcomes at each stage, the classifier is called a binary decision tree classifier (BDT).

Figure 11.5.1 shows the spectral characteristics of ground truth data for nine classes and the corresponding decision tree classifier to classify the nine classes using their spectral characteristics.

Generally a group of classes will be classified into two groups with the highest separability with respect to a feature.

Features often used are as follows.
(1) Spectral values
(2) An index which is computed from spectral values. For example, the vegetation index is a popular indices.
(3) any arithmetic value such as addition, subtraction or ratioing.
(4) Principal components.

The advantages of the decision tree classifier are that computing time is less than the maximum likelihood classifier and by comparison the statistical errors are avoided. However the disadvantage is that the accuracy depends fully on the design of the decision tree and the selected features.

Figure 11.5.2 shows an example of classification with a decision tree classifier.


Copyright © 1996 Japan Association of Remote Sensing All rights reserved