Local chiken breeds valorization by image analysis application on eggs produced in organic system

The evaluation of food color is a psychological problem of sensitivity to attractiveness or to irritation. In the case of egg yolk color there exists an "optimal-yellow”, which must be "appetizingly pretty," as determined by the "corruptible human eye". The yolk color is affected by genetics, housing and feeding system. Actually, for yolk measurement, the scale consists of 15 different color stripes ranging from light yellow to dark orange and red according to the Yolk Color Fan(Roche)scale. Consumer acceptability thus depends on visual impression, but the human eye is not very sensitive to the darker shades of yellow. To solve this problem, the aim of this study was to assess the pattern of color pigmentation of the egg yolk of Siciliana, Livornese and Lohmann White hens reared in organic system, using red green blue (RGB) image, for a future standardized technique with lowering human error by individual visual perception. For the trial, 63 eggs were sampled from 3 groups (21 eggs/group) of chicken breeds reared in organic system: Siciliana (S), Livorno (L) and Lohman White (LW). The individual egg yolks were placed on Petri dishes (50mm diameter) and photographed in a measurement chamber, with a camera for high-resolution data acquisition (16 million colors) by using an E-eye (Iris Visual Analyzer 400-Alpha MOS). The application of the software available in the instrument (Alphasoft, version 14.0) allowed to group color spectra in range of 16 bit for each coordinates RGB obtaining 4096 variables shown as histograms. To evaluate the ability of the E-eye in discriminating the different egg-producing breed, data collected on the samples of each group were processed by PCA. A selection of the most discriminant variables has been performed in order to improve the separation between samples. Results showed that, for S group, greater color homogeneity described by the predominance of a lower number of bars (colors) was seen (5 codes color); on the contrary, the number of bars increased passing from L (7 codes color) to LW group (11 codes color). The PCA analysis explained 99.53% of the total variance (98.61 for PC1 and 0.93% for PC2). Considering the locations of products on the surface (PCA score) was possible to note that S and L samples were quite grouped in a cluster, whereas LW samples were clearly differentiated from S and L, but divided in two groups mainly as a function of PC1. Different direction of vectors (PCA loadings), shows which variables (colors) were involved in the appearance variations among samples. Variables “colors 2144 and 2400" which describe the strongest yellow intensity affected mainly the position of S samples, on the contrary, the darkness variable “color 2128", was opposite and characterized L samples. Our results, which described yolk eggs colors and identified the 3 breeds, suggest that image processing can be applied to extract RGB image of yolk color, which could be used to develop the model of color recognition.