Database Reference
In-Depth Information
FIGURE 15.10
Modis data processing.
15.4.5.2 PCA-Based Feature Space Representation
For feature space representation, the first PCA was applied for data visualization,
exploration of data clustering in multisensor space, and to establish sensor correla-
tions. The objective of the PCA exploration was to establish whether or not simple
classes exist in the feature space and to see whether the data clusters could be found
by investigation, before the self organizing knowledge recommendation. The PCA
method consists of expressing the response vectors in terms of a linear combination
Search WWH ::




Custom Search