resampleSignature

Read the spectral signature of vegetation from the ECOSTRESS spectral library, to use as the reference spectrum.

lib = readEcostressSig("vegetation.tree.tsuga.canadensis.vswir.tsca-1-47.ucsb.asd.spectrum.txt");

Load a hyperspectral image into the workspace.

hcube = hypercube("paviaU.dat");

Colorize the hyperspectral image to get an RGB image.

rgbImg = colorize(hcube,Method="rgb",ContrastStretching=true);

Compute the distance scores of the spectrum of the hyperspectral image pixels with respect to the reference spectrum.

scoreBeforeResampling = spectralMatch(lib,hcube);

Resample the reference spectrum to align with the wavelengths of the hyperspectral image.

resampledRef = resampleSignature(lib,hcube.Wavelength);

Compute the distance scores of the spectrum of the hyperspectral image pixels with respect to the resampled reference spectrum.

scoreAfterResampling = spectralMatch(resampledRef,hcube);

Visualize the RGB image of the hyperspectral image, the distance score map before resampling the reference spectrum, and the distance score map after resampling the reference spectrum. The pixels with low distance scores are stronger matches to the reference spectrum, and are more likely to belong to the vegetation region. Observe that the distance scores after resampling the reference spectrum are more accurate, as regions that do not have any vegetation cover have a significantly higher distance score after resampling the reference spectrum.

figure
tiledlayout(3,1)
nexttile
imshow(rgbImg)
title("RGB Image")
nexttile
imagesc(scoreBeforeResampling)
title("Spectral match before resampling")
axis image off
colormap(parula(5))
colorbar
nexttile
imagesc(scoreAfterResampling)
title("Spectral match after resampling")
axis image off
colormap(parula(5))
colorbar

Load a .mat file that contains the reflectance and the wavelength values of a spectrum into the workspace. Check the length of the reflectance spectrum. Note that the spectrum has reflectance values for 198 wavelengths. Check the range of wavelengths in the spectrum.

load spectralData.mat reflectance wavelength
length(reflectance)

ans = 198

min(wavelength)

ans = 399.3700

max(wavelength)

ans = 2.4572e+03

Load a hyperspectral image into the workspace. Note that the hyperspectral image has 103 spectral bands. Check the range of wavelengths in the hyperspectral image. Observe that all the wavelength range of the hyperspectral image lies within the wavelength range of the reflectance spectrum. Thus, all wavelengths of the hyperspectral image will comprise the common wavelengths between the hyperspectral image and the reflectance spectrum.

hcube = hypercube("paviaU.dat")

hcube = 
  hypercube with properties:

      DataCube: "[610x340x103 double]"
    Wavelength: [103x1 double]
      Metadata: [1x1 struct]

min(hcube.Wavelength)

ans = 430

max(hcube.Wavelength)

ans = 838

Resample the reflectance spectrum to align with the wavelengths of the hyperspectral image.

[resampledRef,commonWavelength,index] = resampleSignature(reflectance,hcube.Wavelength,wavelength);

Check the length of the reflectance spectrum and the range of the common wavelengths in nanometer. Observe that the resampled spectrum has reflectance values of 103 wavelengths, and the range of the common wavelengths is the same as the wavelength range of the hyperspectral image. Thus, the common wavelengths between the hyperspectral image and the reflectance spectrum comprise all wavelengths in the hyperspectral image. You can confirm this by inspecting the index output, which provides the indices of the first and last common wavelengths within the wavelength range of the hyperspectral image.

length(resampledRef)

ans = 103

min(commonWavelength)*10^3

ans = 430

max(commonWavelength)*10^3

ans = 838

index

index = 1×2

     1   103

Find the indices in the reflectance spectrum of the wavelengths closest to the first and last common wavelengths. Observe that the common wavelength range of the reflectance spectrum consists of only 45 wavelengths. As such, the resampleSignature function uses interpolation to estimate resampled reflectance values for all 103 wavelengths in the common wavelength range.

[~,idxMin] = min(abs(wavelength-commonWavelength(1)*10^3))

idxMin = 4

[~,idxMax] = min(abs(wavelength-commonWavelength(end)*10^3))

idxMax = 48