10 October 2018 8 9K Report

I have a 3D dataset structured as : (x,y) are pixels of 2D image, z are spectra that are acquired on each position (x,y). The target is to cluster pixels on (x,y) using the distance of their spectra (in z).

I have already applied the tSNE and k-means on this dataset using the correlation between each spectra, however I am interested to know if there are new distance metrics or clustering algorithms that could be very useful for clustering spectral data to give it a try.

Similar questions and discussions