Analysis of Meat Color Change using Computer Vision

Sensory parameters are crucial for making a purchase decision in meat products. Thus, consumers will guide their choice based on their color. They will seek cherry red meat; when the meat turns brown due to the myoglobin's oxidation, the product is no longer desired. Therefore, the food industry must have a system that could be effective and give just-in-time information regarding changes in color to maintain the quality during the shelf life that consumers expect. This research aims to present a methodology based on computer vision to analyze the change of color in meat. We used images taken of different beef cuts and tested on different days. The Euclidean distance on the average of colors could be used. However, the method proposed in this study is the use of Kullback Leibler divergence, which takes the meat not only at one color point but as a cloud of points. The results were obtained with the Kullback Leibler divergence demonstrated that it is possible to calculate differences in meat images when passing the days. The practical application for this type of analysis would be in the retail industry in order to give just in time information about the quality of meat.