Classification of CARS spectral phase retrieval combined with principal component analysis

Abstract

Coherent anti-Stokes Raman scattering spectroscopy suffers from spectral distortion and low signal to noise ratios due to the influence of nonresonant background. To make the spectral information useful, phase retrieval was used through the Kramers-Kronig relation, one of the ways to reduce the constraints due to nonresonant background. By applying this, a comparative analysis was performed on 4 marine oils having a high degree of unsaturation. Principal component analysis was applied to classify samples that were difficult to distinguish only with the intensity of Raman bands in the spectral data and obtained new variables that maximize the data variance. The construction of the scores plot shows that different samples are separated from each other and that the same samples are grouped.