Abstract
Remotely sensed multispectral images are analyzed by the methods of maximum likelihood, Euclidean minimum distance and correlation etc. But a new fast analyzing method is searched because the processing speed of the conventional is slow.
This paper describes the multispectral image processing by means of a binary decision tree. To appreciate the ability of the image analysis, the efficiency of the binary decision tree is compared with the efficiency of the maximum likelihood method under the same conditions.
As a result, the processing speed of the binary decision tree was shown to be twenty times as fast as that of the maximum likelihood with a nine-classes six-features classification. Moreover, the processing time of the binary decision tree method does not increase even if the number of the features increses. This property is another strong point of the binary decision tree. On the contrary, the processing time of the maximum likelihood method is in proportion to the square of the number of selected features. On the other hand, the classification accuracy of the binary decision tree was about equal to that of the maximum likelihood method.
