Determination of pork marbling based on image texture analysis

Authors: Huang, H, L Liu, M Ngadi
Description: The requirement for online marbling assessment in pork industry has encouraged investigation on efficient techniques to assess pork marbling levels in a rapid, accurate, and objective way. For this purpose, an efficient and effective method was developed for determining pork marbling scores based on image texture analysis. In this study, a total of 53 pork loin chops were collected and their marbling scores were assessed subjectively by trained employees of the plant using the NPPC marbling standards. Digital RGB (red-green-blue) images were captured for all pork samples and the loin area was segmented as the ROI (region of interest). Image texture features were extracted from ROI using an improved GLCM (gray-level co-occurrence matrix) which allowed extracting image texture features from irregular-shaped ROI without the influence of non-ROI area. A support vector machine (SVM) model was established to predict the marbling scores based on the extracted image texture features. The high correlation coefficients of calibration and validation (0.81 and 0.80, respectively) demonstrated the potential of the proposed method for pork marbling determination.
Keywords: pork, marbling, gray-level co-occurrence matrix (glcm), support vector machine (svm)
Technical field: technical_fields_app4
Session name: Food and bioprocess engineering
Date: 2012
Identifier: CSBE12084
Coverage: Canada

Search a publication

Search for:
Find entries that have:
Publication type: