How to apply the outlines defined by superpixels to a hyperspectral image to calculate the mean spectrum of each superpixel?

I have applied the function superpixels to extract the outlines of several pixels wich have similar colour within an RGB image. Now I want to apply these outlines to the original hypercube to obtain the mean spectrum for each of these superpixels. Can you help me?

Answer 1

[Parth Parikh]

Hi Laura,

Here is the code you can try:

Suppose you have number of labels (N) and label matrix (L) from superpixels function.

[rows, cols, channels] = size(hypercube);
hypercube = reshape(hypercube, [rows*cols channels]);

idx = label2idx(L); 

outputImg = zeros(rows*cols, channels);
for labelVal = 1:N
    idxs = idx{labelVal};
    outputImg(idxs, :) = repmat(mean(hypercube(idxs,:)), length(idxs),1);
end
outputImg = reshape(outputImg, [rows, cols, channels]);

I am assuming instead of taking mean in spatial dimension, you would like to take a mean of spectral dimension.

If you would like to explore more about the hyperspectral domain, kindly go through this:

Hyperspectral Image Processing - MATLAB & Simulink (mathworks.com)

Image Processing Toolbox Hyperspectral Imaging Library - File Exchange - MATLAB Central (mathworks.com)

[Laura]

Thank you so much!