Development of Noninvasive Classification Methods for Different Roasting Degrees of Coffee Beans Using Hyperspectral Imaging

Advances in Machine Learning and Hyperspectral Imaging in the Food Supply Chain

Food quality and safety are the essential hot issues of social concern. In recent years, there has been a growing demand for real-time food information, and non-destructive testing is gradually replacing traditional manual sensory testing and chemical analysis methods with lagging and destructive effects and has strong potential for application in the food supply chain. With the maturity and development of computer science and spectroscopic techniques, machine learning and hyperspectral imaging (HSI) have been widely demonstrated as efficient detection techniques that can be applied to rapidly evaluate sensory characteristics and quality attributes of food products nondestructively and efficiently. This paper first briefly described the basic concepts of hyperspectral imaging and machine learning, including the imaging process of HSI, the type of algorithms contained in machine learning, and the data processing flow. Secondly, this paper provided an objective and comprehensive overview of the current applications of machine learning and HSI in the food supply chain for sorting, packaging, transportation, storage, and sales, based on the state-of-art literature from 2017 to 2022. Finally, the potential of the technology is further discussed to provide optimized ideas for practical application.

Analysing the water spectral pattern by near-infrared spectroscopy and chemometrics as a dynamic multidimensional biomarker in preservation: rice germ storage monitoring

Water activity is an important phenomenon not yet explained in terms of water molecular structure. This paper aims to find the relationship between the water activity and water molecular structure of the rice germ, based on its spectral pattern which can be measured using non-destructive technology. Aquaphotomics near-infrared spectroscopy was used to study rice germ stored at different levels of water activity and atmosphere. The findings show that state of the rice germ is governed by the water activity upon storage, which is defined by the structure of water within germ matrix. The structure of water can be described solely by the absorbance spectral pattern at the following absorbance bands: proton hydrates, hydration shells and water vapor (1364, 1375 and 1382 nm), trapped water (1392 nm), free water (1410 nm), hydration water (1425 nm), adsorbed water (1455 nm), non-bonded hydroxyl (1436 nm) and bound water (1520 nm).

Monitoring of the residual moisture content in finished textiles during converting by NIR hyperspectral imaging

Hyperspectral imaging was used for large-scale monitoring of the residual moisture in wide textile webs at the end of the drying process that follows their washing or finishing by impregnation in aqueous solutions or dispersions. Such data are essential for optimizing the energy efficiency and the precise control of the drying process. Quantitative analysis of the recorded spectral data was carried out with multivariate regression methods such as the partial least squares (PLS) algorithm. Reference data for calibration of the prediction models were determined by gravimetry. The drying of textile materials from both natural or synthetic fibers possessing different water absorption capacities (cotton, polyamide, polyester), which were partially finished with an optical brightener, was investigated. Moisture contents in the range from 0 to about 12 wt% were considered in the calibration models. For all systems, the root mean square error of prediction (RMSEP) for the residual moisture was found to be about 0.5 wt%, that is, about 1 g/m². In addition to the quantitative determination of the water content, hyperspectral imaging provides detailed information about its spatial distribution across the textile web, which may help to improve the control of the drying process. In particular, it was demonstrated that the developed methods were capable of detecting and visualizing inhomogeneous moisture distributions. Averaging of the individual values of the moisture content predicted from all spectra across the surface of the textile samples resulted in a very close correlation with the corresponding gravimetric reference values. Due to the averaging process, the difference between both values is generally lower than RMSEP even in case of samples with inhomogeneous distribution of the moisture. The high precision and the broad capabilities of the developed analytic methods for in-line monitoring of the moisture content hold the potential for an efficient process control in technical textile converting processes.

Prediction of Sweet Corn Seed Germination Based on Hyperspectral Image Technology and Multivariate Data Regression

Vigor identification in sweet corn seeds is important for seed germination, crop yield, and quality. In this study, hyperspectral image (HSI) technology integrated with germination tests was applied for feature association analysis and germination performance prediction of sweet corn seeds. In this study, 89 sweet corn seeds (73 for training and the other 16 for testing) were studied and hyperspectral imaging at the spectral range of 400-1000 nm was applied as a nondestructive and accurate technique to identify seed vigor. The root length and seedling length which represent the seed vigor were measured, and principal component regression (PCR), partial least squares (PLS), and kernel principal component regression (KPCR) were used to establish the regression relationship between the hyperspectral feature of seeds and the germination results. Specifically, the relevant characteristic band associated with seed vigor based on the highest correlation coefficient (HCC) was constructed for optimal wavelength selection. The hyperspectral data features were selected by genetic algorithm (GA), successive projections algorithm (SPA), and HCC. The results indicated that the hyperspectral data features obtained based on the HCC method have better prediction results on the seedling length and root length than SPA and GA. By comparing the regression results of KPCR, PCR, and PLS, it can be concluded that the hyperspectral method can predict the root length with a correlation coefficient of 0.7805. The prediction results of different feature selection and regression algorithms for the seedling length were up to 0.6074. The results indicated that, based on hyperspectral technology, the prediction of seedling root length was better than that of seed length.

Characterization of Arabica and Robusta Coffees by Ion Mobility Sum Spectrum

Aroma is one of the main characteristics of coffee specimens. Different mixtures of Arabica and Robusta coffees are usually found in the market to offer specific aroma or flavor profiles to consumers. However, the mixed samples or their proportions are not always identified in the product labels. Since the price of Arabica is much higher than that of Robusta, this lack of information is not only an economical issue but a possible fraud to consumers, besides the potential allergic reaction that these mixtures may trigger in some individuals. In this paper, two sample preparation techniques were compared before the analysis of the total volatile organic compounds (VOCs) found in Robusta, Arabica, and in the mixture from both coffee types. The comparison of the signals obtained from the analyses showed that the VOCs concentration levels obtained from the headspace (HS) analyses were clearly higher than those obtained from the pre-concentration step where an adsorbent, an active charcoal strip (ACS + HS), was used. In the second part of this study, the possibility of using the headspace gas-chromatography ion mobility spectrometry (HS-GC-IMS) for the discrimination between Arabica, Robusta, and mixed coffee samples (n = 30) was evaluated. The ion mobility sum spectrum (IMSS) obtained from the analysis of the HS was used in combination with pattern recognition techniques, namely linear discrimination analysis (LDA), as an electronic nose. The identification of individual compounds was not carried out since chromatographic information was not used. This novel approach allowed the correct discrimination (100%) of all of the samples. A characteristic fingerprint for each type of coffee for a fast and easy identification was also developed. In addition, the developed method is ecofriendly, so it is a good alternative to traditional approaches.

Investigating the effects of geographical origin, roasting degree, particle size and brewing method on the physicochemical and spectral properties of Arabica coffee by PCA analysis

Effects of geographical origin of Coffee Arabica beans (Brazil, Colombia and Peru), roasting degree, particle size and different brewing methods (French press, chemex and cold brew) on physicochemical and spectral properties of coffee samples were investigated in this study. Analyses of pH, total phenolic content (TPC) and total antioxidant activity (TA), UV and fluorescence spectroscopy measurements were performed. Principal component analysis (PCA) was used to obtain the most effective parameters on chemical changes. Results indicated that the increase of roasting degree caused decrements in the intensities of UV-spectra because of the degradation of trigonelline and chlorogenic acid molecules while counterpart trend was observed in the fluorescence spectra due to formation of fluorescence Maillard reaction products (MRP) during roasting. French press and cold brew methods caused similar TPC (1873.33-3818.33 and 2648.88-3824.44 μg/mL gallic acid equivalent, respectively) and TA (0.18-0.32 and 0.16-0.27 μmol/mL Trolox equivalent, respectively) values whereas chemex method showed different physicochemical properties (TPC: 1008.88-3543.88 μg/mL gallic acid equivalent and TA: 0.08-0.26 μmol/mL Trolox equivalent). Roasting degree and brewing method-compared to other parameters-were the most discriminating factors on the basis of UV spectra and fluorescence spectra of coffee brew samples, respectively. All roasting degrees could be distinguished with the rate of 71.42% on PC1 and 23.45% on PC2 of total variance according to UV-spectra while chemex and French press-cold brew methods could be differentiated with the rate of 97.24% on PC1 and 1.79% on PC2 of total variance based on fluorescence spectra on PCA score graphs.

Hyperspectral imaging technology combined with deep forest model to identify frost-damaged rice seeds

In recent years, deep learning models have been widely used in the field of hyperspectral imaging. However, the training of deep learning models requires not only a large number of samples, but also the need to set too many hyper-parameters, which is time consuming and laborious for researchers. This study used hyperspectral imaging technology combined with a deep learning model suitable for small-scale sample data sets, deep forests (DF) model, to classify rice seeds with different degrees of frost damage. During the period, three spectral preprocessing methods (Savitzky-Golay first derivative (SG1), standard normal variate (SNV), and multivariate scatter correction (MSC)) were used to process the original spectral data, and three feature extraction algorithms (principal component analysis (PCA), successive projections algorithm (SPA), and neighborhood component analysis (NCA)) were used to extract the characteristic wavelengths. Moreover, DF model and three traditional machine learning models (decision tree (DT), k-nearest neighbor (KNN), and support vector machine (SVM)) were built based on different numbers of sample sets. After multivariate data analysis, it showed that the pretreatment effect of MSC was the most excellent, and the characteristic wavelength extracted by NCA algorithm was the most useful. In addition, the performance of DF model was better than these three traditional classifier models, and it still performed well in small-scale sample set data. Therefore, DF model was chosen as the best classification model. The results of this study show that the DF model maintains good classification performance in small-scale sample set data, and it has a good application prospect in hyperspectral imaging technology.

Emerging Technologies Based on Artificial Intelligence to Assess the Quality and Consumer Preference of Beverages

Beverages is a broad and important category within the food industry, which is comprised of a wide range of sub-categories and types of drinks with different levels of complexity for their manufacturing and quality assessment. Traditional methods to evaluate the quality traits of beverages consist of tedious, time-consuming, and costly techniques, which do not allow researchers to procure results in real-time. Therefore, there is a need to test and implement emerging technologies in order to automate and facilitate those analyses within this industry. This paper aimed to present the most recent publications and trends regarding the use of low-cost, reliable, and accurate, remote or non-contact techniques using robotics, machine learning, computer vision, biometrics and the application of artificial intelligence, as well as to identify the research gaps within the beverage industry. It was found that there is a wide opportunity in the development and use of robotics and biometrics for all types of beverages, but especially for hot and non-alcoholic drinks. Furthermore, there is a lack of knowledge and clarity within the industry, and research about the concepts of artificial intelligence and machine learning, as well as that concerning the correct design and interpretation of modeling related to the lack of inclusion of relevant data, additional to presenting over- or under-fitted models.

Aquaphotomics-From Innovative Knowledge to Integrative Platform in Science and Technology

Aquaphotomics is a young scientific discipline based on innovative knowledge of water molecular network, which as an intrinsic part of every aqueous system is being shaped by all of its components and the properties of the environment. With a high capacity for hydrogen bonding, water molecules are extremely sensitive to any changes the system undergoes. In highly aqueous systems-especially biological-water is the most abundant molecule. Minute changes in system elements or surroundings affect multitude of water molecules, causing rearrangements of water molecular network. Using light of various frequencies as a probe, the specifics of water structure can be extracted from the water spectrum, indirectly providing information about all the internal and external elements influencing the system. The water spectral pattern hence becomes an integrative descriptor of the system state. Aquaphotomics and the new knowledge of water originated from the field of near infrared spectroscopy. This technique resulted in significant findings about water structure-function relationships in various systems contributing to a better understanding of basic life phenomena. From this foundation, aquaphotomics started integration with other disciplines into systematized science from which a variety of applications ensued. This review will present the basics of this emerging science and its technological potential.

Custom Scanning Hyperspectral Imaging System for Biomedical Applications: Modeling, Benchmarking, and Specifications

Prototyping hyperspectral imaging devices in current biomedical optics research requires taking into consideration various issues regarding optics, imaging, and instrumentation. In summary, an ideal imaging system should only be limited by exposure time, but there will be technological limitations (e.g., actuator delay and backlash, network delays, or embedded CPU speed) that should be considered, modeled, and optimized. This can be achieved by constructing a multiparametric model for the imaging system in question. The article describes a rotating-mirror scanning hyperspectral imaging device, its multiparametric model, as well as design and calibration protocols used to achieve its optimal performance. The main objective of the manuscript is to describe the device and review this imaging modality, while showcasing technical caveats, models and benchmarks, in an attempt to simplify and standardize specifications, as well as to incentivize prototyping similar future designs.

A Distributed Parallel Algorithm Based on Low-Rank and Sparse Representation for Anomaly Detection in Hyperspectral Images

Anomaly detection aims to separate anomalous pixels from the background, and has become an important application of remotely sensed hyperspectral image processing. Anomaly detection methods based on low-rank and sparse representation (LRASR) can accurately detect anomalous pixels. However, with the significant volume increase of hyperspectral image repositories, such techniques consume a significant amount of time (mainly due to the massive amount of matrix computations involved). In this paper, we propose a novel distributed parallel algorithm (DPA) by redesigning key operators of LRASR in terms of MapReduce model to accelerate LRASR on cloud computing architectures. Independent computation operators are explored and executed in parallel on Spark. Specifically, we reconstitute the hyperspectral images in an appropriate format for efficient DPA processing, design the optimized storage strategy, and develop a pre-merge mechanism to reduce data transmission. Besides, a repartitioning policy is also proposed to improve DPA’s efficiency. Our experimental results demonstrate that the newly developed DPA achieves very high speedups when accelerating LRASR, in addition to maintaining similar accuracies. Moreover, our proposed DPA is shown to be scalable with the number of computing nodes and capable of processing big hyperspectral images involving massive amounts of data.