Image Analysis, Classification and Change Detection in Remote Sensing, 3rd Edition

8

Unsupervised Classification

Supervised classification o f remote sensing imagery, the subject of the previous

two chapters, involves the use of a tr aining dataset consisting of labeled pixels

representative of each land cover category of interest in an image. We saw

how to use these data to generalize to a complete labeling, or thematic map,

for an e ntire scene. The choice of training areas which adequately represent

the spectral characteristics of each category is very important for supervised

classification, as the quality of the training set has a profound effect o n the

validity of the result. Finding and verifying training areas can be laborious, ...