A Similarity Measure for Two-Dimensional Fluorescent Spectra

Abstract

A similarity measure was developed that can differentiate between two-dimensional fluorescent spectra based on their similarities and differences. The two-dimensional fluorescent spectra are digitalized into matrices. The difference between the two spectra is defined by a difference matrix, whose elements contain the difference of one two-dimensional fluorescent spectrum minus the other. The similarity measure is transformed into hypothesis tests of the similarity and difference between the two spectra. The scalar mean of the difference matrix is used as the statistical variable for the hypothesis test. The Bayesian prior odds ratio was estimated from multiple spectra of the same reference sample. A threshold for the hypothesis test that the spectra are different is proposed. The posterior odds ratio was used to quantify the similarity measure of the two spectra. Two-dimensional fluorescent spectra of Changyu red wine samples were used to demonstrate this method. The results show that this new method can detect differences between the spectra.