Abstract:

In the present study, a hyperspectral imaging technique was investigated for non-destructive determination of
refrigerated time and moisture content in chilled mutton. Totally 160 pieces of spectral information of mutton samples were
collected by hyperspectral image system in the region of 400–1 000 nm. PC-14-LDA method was chosen to establish the
original spectral model for the refrigerated time of chilled mutton. The discriminant rates of refrigerated time in calibration
and validation sets were 99.17% and 100%, and the established model could distinguish the refrigerated time of mutton well.
Secondly, optimal pretreatment method and partial least squares regression (PLSR) were chosen to establish the prediction
model of moisture content in refrigerated mutton. The results showed that the PLSR model behaved for predicting moisture
content in mutton after Savitzky-Golay smoothing. The correlation coefficients of calibration and prediction sets were
0.888 and 0.784, respectively, and root mean square errors of cross-validation were both 0.696. Therefore, near-infrared
hyperspectral imaging technique is feasible for determining the refrigerated time and moisture content in chilled mutton.

Key words: hyperspectral imaging technique, chilled mutton, refrigerated time, moisture content, non-destructive detection