Development of multivariate regression models for real-time monitoring of physicochemical changes of foodstuffs during drying process using Vis-NIR optical imaging

干燥是应用最广泛的食品长期保存方法。目前使用的干燥方法主要导致产品质量的低保留率。多酚类、类胡萝卜素和维生素是对热非常敏感的一些最重要的营养物质。此外，在热加工过程中，物理属性发生了严重的变化。为了优化工艺，充分了解产品（及其组件）在整个过程中发生的动态变化是必不可少的，这只有在可靠的非侵入性质量检测系统可用的情况下才有可能。为此，光学技术作为一种非破坏性、非接触式和快速的工具，用于监测暴露在干燥过程中的食品，在过去的几十年里获得了极大的兴趣。本研究拟基于400-1700 nm光谱范围内的高光谱和四种多光谱成像技术（即激光后向散射、生物斑点、滤光片和led成像技术）获取的数据，建立、评估和比较多元回归模型，以预测苹果的总多酚、总类胡萝卜素、维生素C（抗坏血酸）、可溶性固形物含量（SSC）、水分含量、收缩率、再水化和硬度。热风干燥过程中的土豆和胡萝卜（两个品种和成熟度）。在干燥过程中食品光学监测领域的大多数研究都局限于水分含量的测量。在干燥过程中，多酚、类胡萝卜素和维生素（特别是维生素C）的光学测量方面的知识严重缺乏，特别是在非侵入性实时测量设备和方案的发展方面。此外，之前的研究主要集中在高光谱成像技术的应用上，而当前的研究将探讨用不同的多光谱成像技术替代高光谱成像的可能性，目的是提供符合农业4.0和工业4.0原则的智能干燥系统。生物散斑成像是一种多光谱成像技术，在干燥过程中具有很大的应用潜力，本文将首次对其适用性进行研究。这项研究的结果将大大有助于更深入地了解光学技术及其在干燥过程实时观察中的潜在用途。开发的方法和设置也将为食品加工和产品质量驱动的过程控制领域的更广泛应用开辟新的可能性。

Drying is the most widely used method of long-term foodstuffs preservation. Currently employed drying methods result predominantly in a low retention of product quality. Polyphenols, carotenoids, and vitamins are some of the most important nutritious substances which are very sensitive to heat. Furthermore, physical attributes undergo severe changes during the thermal processing. For the optimization of the process, it is intrinsic to fully understand the dynamic changes of a product (and its components) undergoing throughout the process, which is only possible if reliable non-invasive quality inspection systems are available. To this end, optical techniques as non-destructive, non-contact, and rapid tools for monitoring of foodstuffs exposed to the drying process have gained a significant interest over the last decades. The current study intends to develop, evaluate, and compare multivariate regression models based on the data acquired using a hyperspectral and four multispectral imaging techniques (i.e. laser-light backscattering, biospeckle, Filter- and LED-based imaging techniques) in the spectrum range of 400-1700 nm with the aim of prediction of total polyphenols, total carotenoids, vitamin C (ascorbic acid), soluble solid contents (SSC), moisture content, shrinkage, rehydration, and firmness of apple, potato, and carrot (two varieties and maturities each) during a hot-air drying process. Most research in the field of optical monitoring of foodstuffs exposed to drying process is limited to the measurement of moisture content. There is a severe lack of knowledge in optical measurement of polyphenols, carotenoids, and vitamins (particularly vitamin C) during the drying process, particularly with a view of the development of non-invasive real-time measurement devices and protocols. Furthermore, previous studies have mainly focused on the application of the hyperspectral imaging technique, whilst the current research will investigate the possible replacement of hyperspectral imaging with different multispectral imaging techniques with the aim of delivering a smart drying system which is in line with the principles of Agriculture 4.0 and Industry 4.0. Biospeckle imaging, one of the multispectral imaging techniques which has a great potential for the inclusion in drying process, will for the first time be studied for its appropriateness. The results of this study will significantly contribute to a deeper understanding of optical techniques and their potential use in real-time observation of drying processes. The developed methods and set-ups will also open new possibilities for wider application across the field of food processing and product quality driven process control.