Is it possible to remove 2-3 endmember spectral lines from a spectral plot, and how?

I plotted a spectral plot of 9 endmembers, from this I need to remove spectral lines 2, 4 and 5 or select other lines.

.

[Shivam]

Could you attach the script for me to debug?

[Image Analyst]

Then just don't plot them. Or do you want the user to be able to turn them on or off by checking checkboxes? If so, just store the handles for the plot and delete the handle if it's unchecked and call plot again if it's checked.

[Anju]

These are the endmembers of a hyperspectral image. By using an endmember extraction algorithm like nfinder, PPI, etc., we extracted the endmembers. It is unable to select the endmembers according to our requirements during the plotting process. The spectral plot shown above is the output generated. Is it possible to remove 2 or 3 lines from this?

[Shivam]

If you can provide requirements which decides the spectral plot along with the script and data you are using, we would be able to assist.

Answer 1

[Vinayak Choyyan]

Hi Anju,

I assume your code currently looks something like this

hcube = hypercube('paviaU.hdr');% your data here
endmembers = fippi(hcube.DataCube,8);%find 8 end member signatures using fippi
legendNames=["Endmember 1","Endmember 2","Endmember 3","Endmember 4","Endmember 5","Endmember 6","Endmember 7","Endmember 8"];

figure
plot(endmembers)%plot the 8 end member signatures. 
xlabel('Band Number')
ylabel('Pixel Values')
ylim([0 9000])
legend(legendNames)

If your goal is the just remove a few predetermined endmember signatures, then you can simply delete them from the signature array like so:

hcube = hypercube('paviaU.hdr');% your data here
endmembers = fippi(hcube.DataCube,8);%find 8 end member signatures using fippi
legendNames=["Endmember 1","Endmember 2","Endmember 3","Endmember 4","Endmember 5","Endmember 6","Endmember 7","Endmember 8"];

%remove predetermined signatures, here signature 2, 4 5
signaturesForRemoval=[2,4,5];
endmembers(:,signaturesForRemoval)=[];
legendNames(signaturesForRemoval)=[];

figure
plot(endmembers)%plot the remaining end member signatures. 
xlabel('Band Number')
ylabel('Pixel Values')
ylim([0 9000])
legend(legendNames)

Hope this helps. You will need the Hyperspectral Imaging Library for Image Processing Toolbox to use functions like hypercube and fippi I used in the above example.

[Anju]

Yes, this is very helpful for me. Thank you very much.