Quick Detection of Beef Adulteration Using Hyperspectral Imaging Technology Combined with Linear Regression Algorithm

Rapid and non-destructive detection of beef adulteration was investigated by hyperspectral imaging technology(900~1700 nm)combined with linear regression algorithm. Beef sample were adulterated with minced chicken in the range of 2%~98%(w/w)at 2% intervals. By collecting hyperspectral images of adulterated samples and extracting spectral data,the quantitative models were established by partial least square regression(PLSR)and multiple linear regression(MLR). To reduce the high dimensionality collinearity of hyperspectral data and improve model efficiency,several optimal wavelengths were selected by using PLS-β,stepwise and successive projection algorithm(SPA)respectively to simplify the models. The results showed that the MLR model combined with SPA showed the better predictive performance for beef adulteration,indicated the coefficient of determination(R_C²)of 0.99 and root mean square error(RMSEC)of 3.23% in calibration set and R_P²,RMSEP and residual predictive deviation(RPD)of 0.97,5.31%,6.82 in prediction set,respectively. The whole results demonstrated that it was feasible to conduct rapid,non-destructive and quantitative detection of adulterated beef by using hyperspectral imaging technology combined with linear regression algorithm.