An overview of non-destructive technologies for postharvest quality assessment in horticultural crops

Abstract

Artificial intelligence and machine vision are increasingly popular within food supply chains for automated decision making in quality grading and disease identification. There are many types of data that these models can be trained on, and choosing which information is needed is a critical factor in minimising both food loss and cost, while maximising the impact on food quality. Non-destructive technologies give information about crop phenotypes (e.g. external colour, oil content, sweetness) without damaging the crop, allowing a greater and more representative proportion the stored food to be analysed. These non-destructive technologies use different methods to analyse the product, each with different intrinsic capabilities and limitations. Therefore, choosing which technology is most appropriate for each application is a complex and costly decision. This mini-review summarises the physical and chemical basis of how some popular non-destructive technologies function, and how these different methods give unique advantages and limitations. The most popular technologies summarised include Red-Green-Blue (RGB) imaging, visible and near-infrared spectroscopy, and vibrometry. We also review technologies that are growing in popularity, including X-ray imaging, ultraviolet spectroscopy, and magnetic resonance imaging.