Measuring colour of vine tomatoes using hyperspectral imaging

As consumers buy with their eyes, colour is considered one of the most important quality parameters of food products. Traditionally, this is defined by human inspection, or measured using a colorimeter or a spectrophotometer. As the first is subjective and prone to factors like fatigue, this is not ideal for industrial use. The second only measures a small area of the food product, making it difficult to get a clear overview of the colour of the whole sample. To overcome these limitations, hyperspectral imaging has been used in this research to measure the postharvest colour of vine tomatoes. Two methods to calculate the colour based on hyperspectral images are compared. The first is the use of a direct method to calculate the colour from the spectra in terms of CIELab-values, while the second method is a soft modelling approach involving multivariate statistics. The soft modelling method was found to achieve the best results (R2L* = 0.86; R2a* = 0.93; R2b* = 0.42, R2Hue = 0.95, R2Chroma = 0.51), but its applicability is limited to the range of products on which the models have been trained. The direct method is more generally applicable, but was found to lack robustness against intensity variations due to the curvature and glossiness of the tomatoes.