Spectral Angle Mapper(SAM)

Spectral angle mapper(SAM) Kruse, F.A. et al. (1993) ‘The Spectral Image Processing System (SIPS) Interactive Visualization and Analysis of Imaging Spectrometer Data’. is an algorithm well suited to determine the similarity between two separate measurements. It considers features important in LIBS so which peaks are present and in which proportions, disregarding the overall intensity. That makes SAM a great fit for LIBS imaging classification problems. It is computationally very simple and cheap so it can be done easily with huge number of measurement points.

Mathematically it considers the measurement data as a vector with N-dimensional space, where N is the number of wavelength pixels (or peaks or features) of a spectrum, and calculates a similarity between measurements as the angle between these vectors. Then a small angle means a similar spectrum, or similar peaks/features.

Potential drawback to be aware of when using SAM is that it is very sensitive to spectral shift, so if you are comparing data from multiple different measurements do check that you dont have spectral shift or that the data are wavelength calibrated. And be aware that temperature change can cause spectral shift in spectrometers.