Using Cluster Analysis to Improve the Selection of Training Statistics in Classifying Remotely Sensed Data
Abstract
A methodology for using cluster analysis to merge unsupervised classes and supervised training fields together for use in the classification process is described. The technique combines the advantages of the two claSSificatIOn approaches while minimizing the disadvantages. Discriminant analySIS IS used to test the 9uahty of the mergmg process while discrete multivariate analysis techniques are used to assess the accuracy. The results mdlcate that higher accuraCies can be achieved using this proposed approach than from supervised or unsupervised claSSificatIon alone.
Department
Natural Resources and the Environment
Publication Date
9-1-1988
Journal Title
Photogrammetric Engineering and Remote Sensing
Publisher
American Society for Photogrammetry and Remote Sensing
Document Type
Article
Recommended Citation
Chuvieco, E. and R. Congalton. 1988. Using cluster analysis to improve the selection of training statistics in classifying remotely sensed data. Photogrammetric Engineering and Remote Sensing. Vol. 54, No. 9, pp. 1275-1281.
Rights
©1988 American Society for Photogrammetry and Remote Sensing