Abstract:Raman spectroscopy was used to establish a non-destructive and rapid detection method for the relative content of methemoglobin and its reductase activity on the surface of chilled beef.The 900 cm-1-1 700 cm-1 band was selected to compare the original spectrum with different spectrum-preprocessing methods.The results showed that the methemoglobin spectrum exhibited a better calibration and prediction model after normalization pretreatment，reductase activity convolution and smoothing treatment.The partial least square（PLS）regression method combined with the weighted β coefficient was used to select the characteristic wavelength.The extraction of the characteristic wavelengths was correlated with the spectral absorption peaks of molecular groups and the PLS regression method model established by the use of characteristic wavelengths showed less variation in related evaluation parameters compared with full-band spectral modeling，indicating that it can be used as a substitute for full-band spectral modeling.PLS regression model of the band model.Raman spectroscopy was therefore feasible for the non-destructive and rapid detection of the relative content of metmyoglobin and reductase activity on the surface of chilled beef.