Spectral identification by singular value decomposition

With the increasing utilization of multi-spectral imaging sensors, automatic identification of spectral signatures would be an invaluable facility. This paper describes a detailed investigation into the suitability of singular value decomposition (SVD) for spectral identification. The ability of SVD to distinguish between several different spectra is assessed, as is its stability in the presence of noise and its capacity to identify combinations of spectra correctly. These results are compared with previously-reported work using neural networks and genetic algorithms.