BANDS

A Spectral Feature Analysis Tool
BANDS is a Hyperspectral data processing module developed by Applied Analysis Inc. to automatically detect and characterize diagnostic spectral features. It detects and characterizes constituent absorption and emission bands to uncover subtle material spectral features. The software uses a high order derivative analysis algorithm to extract band positions, widths, and strengths while suppressing noise.

BANDS assumes that a spectrum contains a superposition of spectral features along with background continuum and sensor noise. It reports band position, width (FWHM), and strength (intensity).

For a more detailed description of BANDS see R. L. Huguenin and J. L. Jones, “Intelligent information extraction from reflectance spectra: Absorption band positions,” J. Geophys. Res. 91, pp. 9585-9598, 1986.

Two versions of BANDS have been developed:
· BANDS with Automatic Parameter Selector (APS) - fully automatic data smoothing
· Manual BANDS - user controlled smoothing

BANDS software is object-oriented and can easily be integrated into hyperspectral classifiers, band selection algorithms, and other algorithms that require a spectrum or time series be broken down into its constituent features.