Prediction of pork fat attributes using NIR Images of frozen and thawed pork

• NIR images of frozen and thawed pork were tested for fat attributes prediction.
• Image texture features were extracted at optimal wavelengths for MLR modeling.
• NIR images of thawed pork could be used to predict IMF content.
• NIR images of frozen and thawed pork could be used to assess marbling scores.

The potential of NIR hyperspectral images of fresh, frozen, and frozen–thawed pork was investigated to quantify intramuscular fat (IMF) content and marbling score (MS) of pork. A Gabor filter which is a Gaussian function-based texture extraction algorithm was applied for image preprocessing after ROI (region of interest) selection. Both raw and Gabor filtered mean spectra of fresh, frozen, and frozen–thawed pork were calculated and their first derivatives at selected optimal wavelengths were used to establish multiple linear regression (MLR) models. The MLR models based on the first derivative of Gabor filtered mean spectra produced best results for both IMF content prediction and marbling score assessment. Models were used to visualize fat distribution in pork loin. The current study therefore demonstrated the potential of using NIR images of frozen–thawed pork to assess IMF content and using frozen and frozen–thawed pork to evaluate MS of pork.