Spectral Angle Map

Features:

Measures the spectral similarity between the spectra of each pixel in the hyperspectral data and the specified reference spectra (seed) by using the spectral angle mapper (SAM) classification algorithm.  A lower score corresponds to stronger spectral similarity. A score of 0 indicates perfect similarity. Usually, a score > 0.1 indicate no similarity between the pixels. A static similarity threshold will have to be applied to recognize similar and dissimilar areas.

  • The user can interactively select a seed area.

  • Shows a spectral angle map between the seed and the rest of the image.

  • Identifies the area of the highest/lowest angle.

  • Quantifies the area correlated with the seed.

Steps:

Draw an area in the Image Display that you would like to use as a reference. This area is called a seed area.

1.      Press the button Spectra Mathematics on the Main Interface under the SPECTRAL ANALYSIS panel.

2.     Select Spectral Angle Map from a dropdown menu.

3.     Select the 1st  spectrum button.

4.     Draw a region of interest and visualize the mean spectrum.

5.     Click Generate to activate the calculation of the map. A new pop-up window Spectral Angle Map appears after the calculation is performed.

6.     Visualize and quantify a Spectral Angle Map.

The mapper has three panels:

a.       Spectrum of the Seed area shows the average spectrum across the selected seed area.

b.       Histogram of the image where each value is a spectral angle score.

c.       A monochromatic image of the map where each pixel value is a spectral angle score.

2.       The lower angle value corresponds to a closer match between the spectrum from the seed and the pixel in the image.

3.       Use a slider on the histogram to set a threshold to visualize the images with the highest or the lowest similarity. For example, to see the regions on the image with the highest similarity to the seed, move the green boundary on the histogram to the left. The modified image of the map will show the highest similarity.

4.       The total area of similar objects with the defined threshold (low values) is shown in the title above the map. 

5.       Press the Reset button to cancel changes and return the spectral angle map image to the original.

Note: For these calculations, only values between the green line and the left edge of the histogram are counted.

References:

Kruse, F.A., A.B. Lefkoff, J.W. Boardman, K.B. Heidebrecht, A.T. Shapiro, P.J. Barloon, and A.F.H. Goetz. “The Spectral Image Processing System (SIPS)—Interactive Visualization and Analysis of Imaging Spectrometer Data.” Remote Sensing of Environment 44, no. 2–3 (May 1993): 145–63. https://doi.org/10.1016/0034-4257(93)90013-N.

Previous
Previous

Spectral Information Divergence Map

Next
Next

Savitzky-Golay Smoothing