Abstract:

A qualitative model for discriminating water-injected meat, gum-injected meat from normal meat was established by
near infrared spectroscopy (NIR) combined with principal component analysis (PCA) and discriminant analysis. The amount of
water injection had an impact on the discrimination accuracy. The overall discrimination accuracy between normal and adulterated
meat was 94.23% when the amount of water injection was 1.25%–20%, and was increased to 96.96% upon water injection at levels
between 3.75% and 20%. The overall discrimination accuracy for all samples in the prediction set was 94.92%. These results show
that NIR combined with PCA and discriminant analysis is feasible to discriminate water-injected meat, gum-injected meat from
normal meat. Quantitative analysis models of water injection and gum injection were established using partial least squares (PLS)
combined with PCA. On the basis of verification, the root mean square errors of prediction (RMSEP) from the two models were
4.01% and 3.87%, respectively, and the correlation coefficients (r) between the predicted values and the actual values were 0.904 2
and 0.912 8, respectively. Therefore, both models have good prediction performance.

Key words: near infrared spectroscopy (NIR), water-injected meat, gum-injected meat, discriminant analysis, partial least squares