pca

This programs converts a cube into principal component space or reverses a cube from principal component space. The principal component analysis (PCA) is performed over all bands of the input cube. The n x n transformation matrix used for this conversion is attached to the output cube as a table, where n is the number of bands of the input cube. Each vertical column in the transformation matrix is an eigenvector.

To run the program, you must provide an input cube, specify an output cube name, and select either Transform or Inverse as your transformation mode of choice

The transformation matrix can be extracted by using the following command:

tabledump from=[cube name].cub name="Transform Matrix"