Nondestructive rapid detection of quality parameters of chilled ham based on visible/near infrared and hyperspectral imaging technology

Aiming at the problems of single quality inspection methods and poor results of traditional meat products, this paper based on visible/near infrared spectra and visible/near infrared hyperspectral technology, took Shuanghui sandwich sausages as the research object, the main indicators of ham during storage were studied at 4℃. 60 ham visible/near infrared spectra and hyperspectral data were collected, and color (L, a*, b*) and total viable count (TVC) of ham were measured at the same time.In order to improve the predictive performance of the model, multivariate scattering correction (MSC), standard normal variable(SNV) and Savozky-Golay smoothing(SG) were used for preprocessing, and the competitive adaptive reweighted sampling (CARS) algorithm was used to select feature variables, then combined with Partial Least Squares Regression (PLSR) to model the full band and characteristic band respectively. The data analysis showed that after CARS extracted the characteristic wavelengths of near-infrared and hyperspectral, the modeling effects are improved. Among them, the validation set correlation coefficients RP of the hyperspectral color(L, a*, b*) model and TVC model were 0.84, 0.94, 0.92 and 0.96, and the validation set prediction errors SEP were 0.90%, 0.38%, 0.26% and 0.56% respectively. For near-infrared spectroscopy, the color(L*, a*, b*) and TVC prediction models had RP of 0.85, 0.90, 0.91 and 0.90, SEP of 1.59%, 0.55%, 0.26% and 0.58% respectively. The results revealed that both visible/near infrared spectroscopy technology and hyperspectral technology can predict the quality parameters of ham very well. Among them, the hyperspectral prediction results were better and the error was smaller.