Rapid determination of minced beef adulteration using hyperspectral reflectance spectroscopy and multivariate methods

Abstract

This work aimed to assess the potential of hyperspectral reflectance spectroscopy for minced beef adulteration. For this, 30 samples of pure minced beef along with 90 samples adulterated beef at different levels were prepared and analyzed. Multivariate methods were used for spectral analysis to classify adulteration types. The best result was obtained by random forest (RF), where the accuracy of classification in prediction set was 87.78%. Considering the redundant information of reflectance spectra, the optimal wavelengths were selected by successive projection algorithm (SPA) to improve robustness. Selected important wavelengths models have a better classification effect than full wavelengths models. The optimal model developed by RF for detecting adulterantion types achieved an accuracy of 96.87% in prediction set with selected wavelengths. Accordingly, hyperspectral reflectance spectroscopy with multivariate methods can provide the accurate and rapid detection of minced beef adulteration.