Use of classification and regression trees (CART) to classify remotely-sensed digital images
Hélio Radke Bittencourt, Robin T. Clarke · 2004
Binary tree-structured rules can be viewed in terms of repeated splits of subsets of the feature space into two descendant subsets, starting from the entire feature space and ending in a partition of the feature space associated with each class. This paper presents a brief introduction to binary decision trees and shows results obtained in the classifying Landsat-TM and AVIRIS digital images.