Spectral Math

Overview

Spectral Math performs mathematical operations on spectra extracted from one or two image regions.

S1 and S2 are generated by drawing regions of interest on the current hyperspectral image.

The mean spectrum of each ROI is used for the calculation.

Click Draw ROI for S1 and draw a region over the image.

The average spectrum from that region becomes S1.

Click Draw ROI for S2 to define a second spectrum.

S2 is required for operations that compare two spectra.

Calculates the vector magnitude of S1.

This provides a measure of overall spectral signal magnitude.

Scales S1 to a normalized intensity range.

Useful when comparing spectral shape independently of absolute signal magnitude.

Subtracts the second spectrum from the first spectrum at each spectral band.

Useful for visualizing spectral differences between two regions.

Divides S1 by S2 band-by-band.

Useful for relative spectral comparisons.

Near-zero values in S2 can produce unstable ratios.

Applies the base-10 logarithm to S1.

Values must be positive for a finite real-valued result.

Calculates the first spectral derivative of S1.

Useful for emphasizing slopes, shoulders, and transitions between spectral features.

Calculates the second spectral derivative of S1.

Useful for emphasizing narrow spectral features and changes in curvature.

Derivative operations can increase the effect of spectral noise.

Measures the similarity in spectral shape between S1 and S2.

Higher correlation indicates more similar spectral profiles.

Calculates the angle between S1 and S2 as vectors in spectral space.

Smaller angles indicate greater spectral similarity.

Measures spectral dissimilarity between S1 and S2.

Larger values generally indicate greater spectral separation.

1. Load a hyperspectral dataset.

3. Select the desired operation.

5. Draw ROI for S2 when the selected operation requires two spectra.

7. Inspect the resulting spectrum or similarity measurement.

Notes

• ROI-derived spectra depend on the current working datacube.
• Preprocessing applied to the cube will therefore affect Spectral Math results.
• Use representative ROIs and avoid mixing unrelated materials within the same ROI.
• Derivative operations are particularly sensitive to noise.
• Ratio calculations can become unstable when S2 contains values near zero.
• Spectral Contrast Explorer