Evaluation of pork colour: prediction of visual sensory quality of meat from instrumental and computer vision methods of colour analysis

M. longissimus dorsi minced pork patties from three dietary treatment groups of DLY (Duroc/Landrace/Yorkshire) cross bred pigs were packaged in polythene bags and placed in a retail refrigerated display cabinet at 5 degreesC+/-1 degreesC, under fluorescent light (1000 lux) for up to 5 days. Each dietary treatment group consisted of pigs (n=7) fed either a low vitamin E diet (80 mg DL-alpha-tocopheryl acetate/kg of feed), supplemental iron (7 g iron (II) sulphate/kg feed) or supplemental vitamin E (200mg DL-alpha-tocopheryl acetate/kg of feed) + supplemental iron). Samples were subjected to visual colour evaluation by a trained sensory panel (n=8) and an untrained panel (n=8) on days 0, 1, 3 and 5. Instrumental Hunter L*, a* and b* values were measured on each day of analysis using a Minolta colorimeter. In addition RGB (red, green and blue) and Hunter L*, a* and b* values were measured using a digital camera. The use of trained and untrained panellists are both relevant in the visual assessment of meat products. In a previous study O'Sullivan, Byrne, and Martens (2003) indicated that the untrained panellist is analogous to the consumer and how they perceive colour changes in meat. However, the trained panellist is useful in the assessment of unfamiliar products and where a greater degree of discrimination is required. The order of oxidation of the experimental treatments was Control (low vitamin E) < iron/vitamin E < Iron. The presentation and measurement of sensory, hedonic and instrumental measurement for a trained and an untrained assessor group highlight the differences in discriminative ability and how the respective instrumental measurements correlate. In general it appears that the instrumental measurements taken with the Digital Camera were more highly correlated to and predictive of the sensory terms as determined by both trained and untrained sensory panellists, particularly for the red, brown and L value descriptors. This was postulated to be due to the fact that the camera took measurements over the entire surface of samples and thus a more representative measurement was taken compared to the colorimeter. (C) 2003 Elsevier Science Ltd. All rights reserved.