Automatic Intra Muscular Fat Analysis on Dry-Cured Ham Slices

摘要

The analysis of intra-muscular fat (IMF) information in dry-cured ham is a very important step in determining its quality and its final target market. This paper presents a fully automatic method for analyzing the IMF content in slices of dry-cured ham. After pre-processing, the approach obtains an initial IMF segmentation using Bias-corrected Fuzzy C-means (BCFCM) segmentation overcoming the inhomogeneous intensity distribution of ham slices. Subsequently, a volumetric IMF estimation model is proposed based on the distance transform of the segmented slices. Finally, a rule-based labelling is used for grading the fat content in order to assess the importance of the features for IMF estimation. Results obtained in a set of 60 slices show a good correlation (0.92) with a ground truth given by standard but more expensive and time consuming techniques, such as the FoodScan analysis.