Application of Bayesian Super-Resolution to Spectroscopic Data for Precise Characterization of Spectral Peak Shape

Abstract The number of data points digitally recorded spectra have been limited by the multichannel detectors employed, which sometimes impedes precise characterization spectral peak shape. Here we describe a methodology to increase as well signal-to-noise (S/N) ratio applying Bayesian super-resolution in analysis spectroscopic data. In our present method, first, hyperparameters for are determined virtual experiment imitating actual experimental data, and precision reconstruction is confirmed calculation errors from ideal values. For validation applied this method Raman spectra. From 200 reference Si substrate with interval about 0.8 cm −1 , 0.01 was successfully achieved promised precision. spectrum, scattering estimated 520.55 (+0.12, −0.09) comparable precisely value reported previous works. can be various kinds analysis, leading increased ability detect slight differences positions shapes.