147
Using machine vision technology to determine pork intramuscular fat percentage
The objective of this study was to test the usefulness of using machine vision technology as a tool to determine intramuscular fat (IMF) percentage in pork loin samples. Loin samples (2.54-cm chop) collected from 2 processing plants were utilized in this experiment for IMF determination. Fresh LM chops were collected and shipped to North Dakota State University for analysis. After arrival, LM chops were trimmed of subcutaneous fat and any connective tissue. Color images of both sides of the LM sample were acquired using machine vision technology which included a camera, lighting system, and computer. Program code was developed at North Dakota State University to segment the background, lean muscle tissue, and IMF. After segmentation, pixels assigned as lean muscle tissue and IMF were counted in order to calculate an image IMF percentage. Subjective marbling scores (NPB, 2011) were called by an experienced grader from each image. Crude fat percentage was calculated using the ether extract method (AOAC, 1990). Image IMF percentage was compared to ether extract values and subjective marbling score. Results show that subjective marbling score had a correlation of 0.79 with ether extract while image IMF had a correlation of 0.56 with ether extract. The correlation between image IMF and subjective marbling score was 0.56. These results indicate that subjective marbling is currently the most accurate method to determine IMF percentage. However, improvement in machine vision technology shows potential of being a tool for IMF determination in the future.
Keywords: machine vision system, marbling, intramuscular fat