Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Nondestructive detection of pork fat content based on hyperspectral spectroscopy

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2021 ASABE Annual International Virtual Meeting  2100302.(doi:10.13031/aim.202100302)
Authors:   Yang Li, Yankun Peng, Yongyu Li, Qibin Zhuang, Qinghui Guo
Keywords:   Pork fat content; Hyperspectral imaging; CARS; PLSR

Abstract. The fat content of pork affects the nutrition and health quality of pork. In order to study the fat content of pork, hyperspectral imaging technology (1000nm-1700nm) was used to obtain hyperspectral data of pork. Based on the original spectra, standard normal transformation (SNV), multivariate scattering correction (MSC) and 1st derivation (1st D) preprocessed spectra, the partial least square regression (PLSR) prediction models of fat content were established, and the results show that the spectral modeling effect based on MSC preprocessing is the best, the correction set correlation coefficient (Rc), correction set root mean square error (RMSEC), prediction set correlation coefficient (Rp) and prediction set root mean square error (RMSEP) are 0.906 , 0.225%, 0.899 and 0.301%,respectively. In order to eliminate irrelevant information and improve model prediction performance, successive projections algorithm (SPA), competitive adaptive reweighed sampling (CARS), variable importance in projection (VIP), Monte Carlo uninformative variables elimination (MCUVE) and mutual information maximization (MIM) algorithms were used on the basis of MSC preprocessing spectra to select the characteristic wavelengths of the spectrum and establish PLSR prediction models. The results show that the modeling effect of CARS is the best, Rc, RMSEC, Rp and RMSEP are 0.962, 0.235%, 0.948 and 0.275%, respectively. The results of this study show that the CARS characteristic wavelengths selection algorithm can effectively reduce the amount of spectral data and improve the prediction performance of the pork fat content prediction model.

(Download PDF)    (Export to EndNotes)