Collect Endmember Spectra

Use Endmember Collection to select endmember spectra from many sources for use in classification and advanced spectral analysis techniques. See Import Spectra for instructions on importing from various sources.

See the following sections to continue:

Endmember Collection Dialog Components

The following input file information is provided in the text at the top of the Endmember Collection dialog:

The Endmember Collection dialog also contains the Endmember Spectra table. This table enables you to view and edit properties associated with any of the spectra listed in the table. The table contains the following columns:

You can also set the default ENVI colors for all the spectra by right-clicking in the Color column title and selecting Assign default colors to all. You can set all the spectra colors to <none> with Assign default colors to undefined, and you can select Reset Colors to reset all the spectra to their original colors. All of the right-click menu options are also available in the Endmember Collection dialog Options menu.

To change an input file with unknown wavelength units to a file containing known wavelength units, select File > Change input File option from the Endmember Collection dialog menu bar.

Manage Endmember Spectra

Use the File menu in the Endmember Collection dialog to save endmember spectra to an ASCII file, as a spectral library, or as an Endmember Collection file. You can also use the File menu to change the input file or exit the dialog.

Plot Spectra

Delete Spectra

Endmember Options

Use the Options menu in the Endmember Collection dialog to edit the endmember names, edit the endmember colors, or suppress backgrounds (using BandMax).

Select Processing Methods

Use the Algorithm menu in the Endmember Collection dialog to select a classification or advanced spectral processing method. The classification methods available are parallelepiped, minimum distance, Mahalanobis distance, maximum likelihood, Spectral Angle Mapper, and binary encoding. The advanced spectral processing methods available are linear spectral unmixing, matched filtering, mixture tuned matched filtering, and spectral feature fitting.

When using the Mahalanobis distance or maximum likelihood classifiers, you can only import the endmember spectra from ROIs or statistic files because these classifications use the endmember covariance statistics. You must also have at least two endmembers.

  1. From the Endmember Collection dialog menu bar, select Algorithm > classification_method.

    You can also start the classification methods from the Classification > Supervised menu and you can start the advanced spectral processing methods from the Spectral Tools menu.

  2. Click Apply. The Algorithm Parameters dialog appears.
  3. Enter the selected algorithm’s required parameters.