Combining Vis-NIR and NIR Spectral Imaging Techniques with Data Fusion for Rapid and Nondestructive Multi-Quality Detection of Cherry Tomatoes

Overview of Deep Learning and Nondestructive Detection Technology for Quality Assessment of Tomatoes

Tomato, as the vegetable queen, is cultivated worldwide due to its rich nutrient content and unique flavor. Nondestructive technology provides efficient and noninvasive solutions for the quality assessment of tomatoes. However, processing the substantial datasets to achieve a robust model and enhance detection performance for nondestructive technology is a great challenge until deep learning is developed. The aim of this paper is to provide a systematical overview of the principles and application for three categories of nondestructive detection techniques based on mechanical characterization, electromagnetic characterization, as well as electrochemical sensors. Tomato quality assessment is analyzed, and the characteristics of different nondestructive techniques are compared. Various data analysis methods based on deep learning are explored and the applications in tomato assessment using nondestructive techniques with deep learning are also summarized. Limitations and future expectations for the quality assessment of the tomato industry by nondestructive techniques along with deep learning are discussed. The ongoing advancements in optical equipment and deep learning methods lead to a promising outlook for the application in the tomato industry and agricultural engineering.

Prediction of the Quality of Anxi Tieguanyin Based on Hyperspectral Detection Technology

Anxi Tieguanyin belongs to the oolong tea category and is one of the top ten most famous teas in China. In this study, hyperspectral imaging (HSI) technology was combined with chemometric methods to achieve the rapid determination of free amino acid and tea polyphenol contents in Tieguanyin tea. Here, the spectral data of Tieguanyin tea samples of four quality grades were obtained via visible near-infrared hyperspectroscopy in the range of 400-1000 nm, and the free amino acid and tea polyphenol contents of the samples were detected. First derivative (1D), normalization (Nor), and Savitzky-Golay (SG) smoothing were utilized to preprocess the original spectrum. The characteristic wavelengths were extracted via principal component analysis (PCA), competitive adaptive reweighted sampling (CARS), and the successive projection algorithm (SPA). The contents of free amino acid and tea polyphenol in Tieguanyin tea were predicted by the back propagation (BP) neural network, partial least squares regression (PLSR), random forest (RF), and support vector machine (SVM). The results revealed that the free amino acid content of the clear-flavoured Tieguanyin was greater than that of the strong-flavoured type, that the tea polyphenol content of the strong-flavoured Tieguanyin was greater than that of the clear-flavoured type, and that the content of the first-grade product was greater than that of the second-grade product. The 1D preprocessing improved the resolution and sensitivity of the spectra. When using CARS, the number of wavelengths for free amino acids and tea polyphenols was reduced to 50 and 70, respectively. The combination of 1D and CARS is conducive to improving the accuracy of late modelling. The 1D-CARS-RF model had the highest accuracy in predicting the free amino acid (RP2 = 0.940, RMSEP = 0.032, and RPD = 4.446) and tea polyphenol contents (RP2 = 0.938, RMSEP = 0.334, and RPD = 4.474). The use of hyperspectral imaging combined with multiple algorithms can be used to achieve the fast and non-destructive prediction of free amino acid and tea polyphenol contents in Tieguanyin tea.

Determination of soluble solids content in tomatoes with different nitrogen levels based on hyperspectral imaging technique

Tomato is sweet and sour with high nutritional value, and soluble solids content (SSC) is an important indicator of tomato flavor. Due to the different mechanisms of nitrogen uptake and assimilation in plants, exogenous supply of different forms of nitrogen will have different effects on the growth, development, and physiological metabolic processes of tomato, thus affecting the tomato flavor. In this paper, hyperspectral imaging (HSI) technique combined with neural network prediction model was used to predict SSC of tomato under different nitrogen treatments. Competitive adaptive reweighed sampling (CARS) and iterative retained information variable (IRIV) were used to extract the feature wavelengths. Based on the characteristic wavelength, the prediction models of tomato SSC are established by custom convolutional neural network (CNN) model that was constructed and optimized. The results showed that the SSC of tomato was negatively correlated with nitrogen fertilizer concentration. For tomatoes treated with different nitrogen concentrations, the residual predictive deviation (RPD) of CARS-CNN and IRIV-parallel convolutional neural networks (PCNN) reached 1.64 and 1.66, both more than 1.6, indicating good model prediction. This study provides technical support for future online nondestructive testing of tomato quality. PRACTICAL APPLICATION: The CARS-CNN and IRIV-PCNN were the best data processing model. Four customized convolutional neural networks were used for predictive modeling. The CNN model provides more accurate results than conventional methods.

Nondestructive Determination of Epicarp Hardness of Passion Fruit Using Near-Infrared Spectroscopy during Storage

The hardness of passion fruit is a critical feature to consider when determining maturity during post-harvest storage. The capacity of near-infrared diffuse reflectance spectroscopy (NIRS) for non-destructive detection of outer and inner hardness of passion fruit epicarp was investigated in this work. The passion fruits' spectra were obtained using a near-infrared spectrometer with a wavelength range of 10,000-4000 cm-1. The hardness of passion fruit's outer epicarp (F1) and inner epicarp (F2) was then measured using a texture analyzer. Moving average (MA) and mean-centering (MC) techniques were used to preprocess the collected spectral data. Competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA), and uninformative variable elimination (UVE) were used to pick feature wavelengths. Grid-search-optimized random forest (Grids-RF) models and genetic-algorithm-optimized support vector regression (GA-SVR) models were created as part of the modeling process. After MC preprocessing and CARS selection, MC-CARS-Grids-RF model with 7 feature wavelengths had the greatest prediction ability for F1. The mean square error of prediction set (RMSEP) was 0.166 gN. Similarly, following MA preprocessing, the MA-Grids-RF model displayed the greatest predictive performance for F2, with an RMSEP of 0.101 gN. When compared to models produced using the original spectra, the R2P for models formed after preprocessing and wavelength selection improved. The findings showed that near-infrared spectroscopy may predict the hardness of passion fruit epicarp, which can be used to identify quality during post-harvest storage.

Optimisation of an Industrial Optical Sorter of Legumes for Gluten-Free Production Using Hyperspectral Imaging Techniques

The market demand for gluten-free food is increasing due to the growing gluten sensitivity and coeliac disease (CD) in the population. The market requires grass-free cereals to produce gluten-free food. This requires sorting methods that guarantee the perfect separation of gluten contaminants from the legumes. The objective of the research was the development of an optical sorting system based on hyperspectral image processing, capable of identifying the spectral characteristics of the products under investigation to obtain a statistical classifier capable of enabling the total elimination of contaminants. The construction of the statistical classifier yielded excellent results, with a 100% correct classification rate of the contaminants. Tests conducted subsequently on an industrial optical sorter validated the result of the preliminary tests. In fact, the application of the developed classifier was able to correctly select the contaminants from the mass of legumes with a correct classification percentage of 100%. A small proportion of legumes was misclassified as contaminants, but this did not affect the scope of the work. Further studies will aim to reduce even this small share of waste with investigations into optimising the seed transport systems of the optical sorter.