Near-infrared Transmission Spectroscopy Research Articles

Design of an AI-based egg fertility detection system for incubators

The conventional approach to determine the fertility of chicken eggs, called candling, is subjective, time-consuming, and ineffective and thereby results in low hatching rates and financial losses. As a result, a more reliable, accurate, and sustainable system that uses artificial intelligence and automated systems to improve the process of assessing egg fertility for incubation. The system proposes capturing of images of eggs at the early stages of incubation by a camera and sent to a cloud server, where a Convolutional Neural Network (CNN) analyses and classifies the data to determine the fertility status of the eggs. The system aims to increase the quality and hatching rate of chicken eggs, lower labour expenses and human error, and boost the poultry industry's sustainability and profitability. The framework of the system is based on the current poultry egg fertility assessment methods in Zimbabwe, which include candling, infrared thermography, ultrasound, heart rate detection, oxygen flux detection, visible or near-infrared transmittance spectroscopy, and thermal imaging. The system is expected to perform better than existing systems in terms of accuracy, speed, and cost-effectiveness. The experimental results show that, the method's average identification time is 0.210 seconds, and its identification accuracy is 98.40 percent. These results demonstrate the viability of the suggested AI-based method for automatically determining an egg's fertility without human intervention, greatly increasing hatcheries' and related businesses' productivity.

Quantification of API content in pharmaceutical tablets within milliseconds by time-stretch near-infrared transmission spectroscopy

We explored the feasibility of high-speed and high-accuracy quantification of active pharmaceutical ingredient (API) content in tablet products by near-infrared (NIR) spectroscopy to improve the reliability of pharmaceuticals. For this purpose, we employed a high-power NIR time-stretch transmission spectrometer recently developed by us. By using this transmission spectrometer with a multivariate calibration model, we demonstrated the ability to quantify API content with a short measurement time of 3.9 ms per tablet for model pharmaceuticals. For the model tablet, the quantification ability of our spectrometer was comparable to that achieved by a commonly used Fourier-transform NIR (FT-NIR) spectrometer with a measurement time of several seconds. We also confirmed that the effect of irradiating tablets with the NIR pulses used in our spectrometer was negligible.

The GAPS Programme at TNG

Context. The atmospheric characterisation of hot and warm Neptune-size exoplanets is challenging mainly due to their relatively small radius and atmospheric scale height, which reduce the amplitude of atmospheric spectral features. The warm-Neptune HAT-P-11 b is a remarkable target for atmospheric characterisation because of the large brightness of its host star (V = 9.46 mag; H = 7.13 mag). Aims. The aims of this work are to review the main physical and architectural properties of the HAT-P-11 planetary system, and to probe the presence of eight molecular species in the atmosphere of HAT-P-11 b through near-infrared (NIR) high-resolution transmission spectroscopy. Methods. We reviewed the physical and architectural properties of the HAT-P-11 planetary system by analysing transits and occultations of HAT-P-11 b from the Kepler data set as well as HIRES at Keck archival radial-velocity data. We modelled the latter with Gaussian-process regression and a combined quasi-periodic and squared-exponential kernel to account for stellar variations on both (short-term) rotation and (long-term) activity-cycle timescales. In order to probe the atmospheric composition of HAT-P-11 b, we observed four transits of this target with the NIR GIANO-B at TNG spectrograph and cross-correlated the data with template atmospheric transmission spectra. Results. We find that the long-period radial-velocity signal previously attributed to the HAT-P-11 c planet (P ~ 9.3 yr; Mp sin i ~ 1.6 MJ; e ~ 0.6) is more likely due to the stellar magnetic activity cycle. Nonetheless, the HIPPARCOS-Gaia difference in the proper-motion anomaly suggests that an outer-bound companion might still exist. For HAT-P-11 b, we measure a radius of Rp = 0.4466 ± 0.0059 RJ, a mass of Mp = 0.0787 ± 0.0048 MJ, a bulk density of ρp = 1.172 ± 0.085 g cm−3, and an orbital eccentricity of e = 0.2577−0.0025+0.0033. These values are compatible with those from the literature. Probing its atmosphere, we detect the presence of two molecular species, H2O and NH3, with a S/N of 5.1 and 5.3, and a significance of 3.4 σ and 5.0 σ, respectively. We also tentatively detect the presence of CO2 and CH4, with a S/N of 3.0 and 4.8, and a significance of 3.2 σ and 2.6 σ, respectively. Conclusions. We revisit the HAT-P-11 planetary system, confirm the presence of H2O, and report the detection of NH3 in the atmosphere of HAT-P-11 b, also finding hints for the presence of CO2 and CH4 that need to be confirmed by further observations.

Online Detection of Dry Matter in Potatoes Based on Visible Near-Infrared Transmission Spectroscopy Combined with 1D-CNN

More efficient resource utilization and increased crop utilization rate are needed to address the growing demand for food. The efficient quality testing of key agricultural products such as potatoes, especially the rapid testing of key nutritional indicators, has become an important strategy for ensuring their quality and safety. In this study, visible and near infrared (Vis/NIR) transmittance spectroscopy (600–900 nm) was used for the online analysis of multiple quality parameters in potatoes. The study concentrated on comparing three one-dimensional convolutional neural network (1D-CNN) models, specifically, the fine-tuned DeepSpectra, the fine-tuned 1D-AlexNet, and classic CNN, with UVE-PLS (uninformative variable elimination–partial least squares) models. These models utilized spectral data for the real-time detection of dry matter (DM) content in potatoes. To address the challenges posed by limited data from Vis/NIR, this study strategically implemented data augmentation techniques. This approach significantly enhanced the robustness and generalization capabilities of the models. The 1D-AlexNet and DeepSpectra models achieved 0.934 and 0.913 R2P and 0.0603 and 0.0695 g/100 g RMSEP for DM, respectively. Compared to UVE-PLS, the R2P value improved by 21.31% (0.770 to 0.934) for the 1D-AlexNet model and 18.64% (0.770 to 0.913) for the DeepSpectra model. The RMSEP value was reduced by 47.31% (0.114 to 0.0603) for 1D-AlexNet, and 39.30% (0.114 to 0.0695) for the DeepSpectra model. As a result, this study would be helpful for researching the online Vis/NIR transmission determination of potato DM using deep learning. These results highlighted the immense potential of employing specific spectral features in deep-learning models for a more precise and efficient online assessment of agricultural quality. This advancement provided some insight and reference for further contributing to the evolution of more targeted and efficient quality assessment methods in agricultural products.

Discrimination of male-sterility and male-fertility in Japanese cedar (Cryptomeria japonica) using near-infrared diffuse transmission spectroscopy

The increasing demand for pollen-free seedlings of Japanese cedar (Cryptomeria japonica) has created a need for a simple method to discriminate between male-sterile and male-fertile strobili. The objective of this study was to establish a classification model to quickly and easily distinguish male-sterile and male-fertile strobili in C. japonica using near-infrared (NIR) diffuse transmission spectroscopy. The absorbance spectra of C. japonica were obtained for three different months from December 2022 to February 2023 and preprocessed using three methods: untreated, smoothing, and second derivative. Principal component analysis was applied to the NIR spectra and classification models were built using a support vector machine. The sample collected in January 2023 showed the highest discrimination accuracy of 89.38% with the smoothing preprocessing, which was improved to 89.97% by limiting the wavelengths to the NIR region. Furthermore, discrimination accuracy for independent test data was evaluated by splitting the data into training and testing sets using January 2023 data with smoothing preprocessing. The discrimination accuracy for test data sets was more than 85%, and the misclassification ratio was less than 20% for each sample group. These results indicate the potential of using NIR diffuse transmission spectroscopy to discriminate between male-sterility and fertility in C. japonica.

Near-infrared transmission spectroscopy of HAT-P-18 b with NIRISS: Disentangling planetary and stellar features in the era of JWST

ABSTRACT The JWST Early Release Observations (ERO) included a NIRISS/SOSS (0.6–2.8 μm) transit of the ∼ 850 K Saturn-mass exoplanet HAT-P-18 b. Initial analysis of these data reported detections of water, escaping helium and haze. However, active K dwarfs like HAT-P-18 possess surface heterogeneities – star-spots and faculae – that can complicate the interpretation of transmission spectra, and indeed, a spot-crossing event is present in HAT-P-18 b’s NIRISS/SOSS light curves. Here, we present an extensive reanalysis and interpretation of the JWST ERO transmission spectrum of HAT-P-18 b, as well as HST/WFC3 and Spitzer/IRAC transit observations. We detect H2O (12.5σ), CO2 (7.3σ), a cloud deck (7.4σ), and unocculted star-spots (5.8σ), alongside hints of Na (2.7σ). We do not detect the previously reported CH4 (log CH4 < −6 to 2σ). We obtain excellent agreement between three independent retrieval codes, which find a sub-solar H2O abundance (log H2O ≈ −4.4 ± 0.3). However, the inferred CO2 abundance (log CO2 ≈ −4.8 ± 0.4) is significantly super-solar and requires further investigation into its origin. We also introduce new stellar heterogeneity considerations by fitting for the active regions’ surface gravities – a proxy for the effects of magnetic pressure. Finally, we compare our JWST inferences to those from HST/WFC3 and Spitzer/IRAC. Our results highlight the exceptional promise of simultaneous planetary atmosphere and stellar heterogeneity constraints in the era of JWST and demonstrate that JWST transmission spectra may warrant more complex treatments of the transit light source effect.

ATMOSPHERIX: II- Characterizing exoplanet atmospheres through transmission spectroscopy with SPIRou

ABSTRACT In a companion paper, we introduced a publicly available pipeline to characterize exoplanet atmospheres through high-resolution spectroscopy. In this paper, we use this pipeline to study the biases and degeneracies that arise in atmospheric characterization of exoplanets in near-infrared ground-based transmission spectroscopy. We inject synthetic planetary transits into sequences of SPIRou spectra of the well known M dwarf star Gl 15 A, and study the effects of different assumptions on the retrieval. We focus on (i) mass and radius uncertainties, (ii) non-isothermal vertical profiles, and (iii) identification and retrieval of multiple species. We show that the uncertainties on mass and radius should be accounted for in retrievals and that depth-dependent temperature information can be derived from high-resolution transmission spectroscopy data. Finally, we discuss the impact of selecting wavelength orders in the retrieval and the issues that arise when trying to identify a single species in a multispecies atmospheric model. This analysis allows us to understand better the results obtained through transmission spectroscopy and their limitations in preparation to the analysis of actual SPIRou data.

Nondestructive Detecting Maturity of Pineapples Based on Visible and Near-Infrared Transmittance Spectroscopy Coupled with Machine Learning Methodologies

Pineapple is mainly grown in tropical regions and consumed fresh worldwide due to its attractive flavor and health benefits. With increasing global production and trade volume, there is an urgent need for nondestructive techniques for accurate and efficient detection of the internal quality of pineapples. Therefore, this study is dedicated to developing a nondestructive method for real-time determining the internal quality of pineapples by using VIS/NIR transmittance spectroscopy technique and machine learning methodologies. The VIS/NIR transmittance spectrums ranging in 400–1100 nm of total 195 pineapples were collected from a dynamic experimental platform. The maturity grade and soluble solids content (SSC) of individual pineapples were then measured as indicators of internal quality. The qualitative model for discriminating maturity grades of pineapple achieved a high accuracy of 90.8% by the PLSDA model for unknown samples. Meanwhile, the quantitative model for determining SSC also reached a determination coefficient (RP2) of 0.7596 and a root mean square error of prediction (RMSEP) of 0.7879 °Brix by the ANN-PLS model. Overall, high model performance demonstrated that using VIS/NIR transmittance spectroscopy technique coupled with machine learning methodologies could be a feasible method for nondestructive and real-time detection of the internal quality of pineapples.

Real-time detection of ammonium in soil pore water

The development of technologies for continuous measurement of nitrogen forms in the soil is essential for optimizing the application of fertilizers in agriculture and preventing water-resource pollution. However, there is no effective commercial technology available for continuous monitoring of ammonium species in soil pore water. This work investigates an approach for real-time measurement of ammonium in soil water using near-infrared transmission spectroscopy and partial least squares regression (PLSR) for spectral analysis. The PLSR model was trained using soil pore water collected from various soils spiked with ammonium to achieve a wide concentration range. The monitoring approach was then validated through transport experiments in a soil column. The results demonstrated capabilities for real-time tracking of the temporal variation in soil ammonium concentration and potential utilization in agronomical or environmental sensing.

BP neural network to predict shelf life of channel catfish fillets based on near infrared transmittance (NIT) spectroscopy

The objective of this research was to establish shelf-life prediction model of channel catfish fillets by Back-propagation (BP) neural network technology based on near infrared transmittance (NIT). First, freshness prediction model of channel catfish fillets was established based on the chemical analysis data (total volatile basic nitrogen (TVB-N), K value, thiobarbituric acid reactive substance (TBARS) and trimethylamine (TMA)) and NIT spectra (850–1050 nm). The linear correlation coefficient (R2: 0.667–0.887) showed a good performance of the freshness model prediction. Then, BP neural network was applied to establish the shelf-life prediction model of catfish fillets under temperature fluctuation (−6 to −18 °C). The end effective accumulated temperature of frozen catfish fillets was 10,278.4 h °C. The prediction model showed a great stability (above 93 %) and accuracy (above 90 %) as the structure of BP neural network was 4–7–1. Therefore, this study provided a practical basis and technical supports for the establishment of shelf-life prediction model of freshwater fillets by BP neural network based on NIT spectroscopy.

Fast detection of volatile fatty acids in biogas slurry using NIR spectroscopy combined with feature wavelength selection.

To analyze the state of anaerobic digestion (AD), fast detection models of volatile fatty acids (VFAs) were constructed using near-infrared transmission spectroscopy combined with partial least squares regression to measure concentrations of the acetic acid (AA), propionic acid (PA) and total acid (TA) in biogas slurry. CARS-SA-BPSO algorithm was proposed based on competitive adaptive reweighted sampling (CARS) and simulated annealing binary particle swarm optimization algorithm (SA-BPSO) for selecting feature wavelengths of the AA, PA and TA. Regression models were established with the determination coefficient of prediction (Rp2) of 0.989, root mean squared error of prediction (RMSEP) of 0.111 and residual predictive deviation (RPD) of 9.706 for AA; Rp2 of 0.932, RMSEP of 0.116 and RPD of 3.799 for PA; Rp2 of 0.895, RMSEP of 0.689 and RPD of 3.676 for TA. It is sufficient to meet the fast detection needs of the AA and PA concentrations in biogas slurry, and basically meet the measuring demand of the TA concentration. CARS-SA-BPSO effectively improves the performance of the calibration model using sensitive wavelength selections, which provides theoretical support for establishing the spectral quantitative regression model to meet the requirements of practical application.

A Mirage or an Oasis? Water Vapor in the Atmosphere of the Warm Neptune TOI-674 b

We report observations of the recently discovered warm Neptune TOI-674 b (5.25 R ⊕, 23.6 M ⊕) with the Hubble Space Telescope’s Wide Field Camera 3 instrument. TOI-674 b is in the Neptune desert, an observed paucity of Neptune-size exoplanets at short orbital periods. Planets in the desert are thought to have complex evolutionary histories due to photoevaporative mass loss or orbital migration, making identifying the constituents of their atmospheres critical to understanding their origins. We obtained near-infrared transmission spectroscopy of the planet’s atmosphere with the G141 grism. After extracting, detrending, and fitting the spectral light curves to measure the planet’s transmission spectrum, we used the petitRADTRANS atmospheric spectral synthesis code to perform retrievals on the planet’s atmosphere to identify which absorbers are present. These results show moderate evidence for increased absorption at 1.4 μm due to water vapor at 2.9σ (Bayes factor = 15.8), as well as weak evidence for the presence of clouds at 2.2σ (Bayes factor = 4.0). TOI-674 b is a strong candidate for further study to refine the water abundance, which is poorly constrained by our data. We also incorporated new TESS short-cadence optical photometry, as well as Spitzer/IRAC data, and refit the transit parameters for the planet. We find the planet to have the following transit parameters: R p /R * = 0.1135 ± 0.0006, T 0 = 2458544.523792 ± 0.000452 BJD, and P = 1.977198 ± 0.00007 day. These measurements refine the planet radius estimate and improve the orbital ephemerides for future transit spectroscopy observations of this highly intriguing warm Neptune.

Detecting Aflatoxin B1 in Peanuts by Fourier Transform Near-Infrared Transmission and Diffuse Reflection Spectroscopy.

Aflatioxin B1 (AFB1) has been recognized by the International Agency of Research on Cancer as a group 1 carcinogen in animals and humans. A fast, batch, and real-time control and no chemical pollution method was developed for the discrimination and quantification prediction of AFB1-infected peanuts by applying Fourier transform near-infrared (FT-NIR) coupled with chemometrics. Initially, the near-infrared transmission (NIRT) and diffuse reflection (NIRR) modules were applied to collect spectra of the samples. The principal component analysis (PCA) method was employed to extract the characteristic wavelength, followed by different preprocessing methods (seven methods) to build an effective linear discriminant analysis (LDA) classification and partial least squares (PLS) quantification models. The results showed that, for both the NIRT or NIRR modules, the LDA classification models satisfactorily distinguished peanuts infected with AFB1 or from those not infected, with external validation showing a 100% correct identification rate and a 0% misjudgment rate. In addition, combined with the concentration of AFB1 in peanuts determined by enzyme-linked immunoassay assay, the best partial least squares (PLS) models were established, with a combination of the first derivative and the Norris derivative filter smoothing pretreatment (Rc2 = 0.937 and 0.984, RMSECV = 3.92% and 2.22%, RPD = 3.98 and 7.91 for NIRR and NIRT, respectively). The correlation coefficient between the predicted value and the reference value in the external verification was 0.998 and 0.917, respectively. This study highlights that both spectral acquisition modules meet the requirements of online, rapid, and accurate identification of peanut AFB1 infection in the early stages.

Multi-layer Raman chemical mapping to investigate the effect of API particle size and blending shear rate on API domain sizes in pharmaceutical tablets

While macromixing (gross uniformity) has received a lot of attention in pharmaceutical powder blending, micromixing (particularly, particle-level aggregation) has been significantly less studied. This study investigated the impact of active pharmaceutical ingredient (API) particle size (D50: 11, 28, and 70 µm) and blending shear rate (low and high) that was caused by tumbling blending (specifically, a V-blender) on micro-mixing. The effect on micro-mixing (API domain sizes) was assessed in direct compression tablets using high-resolution Raman chemical mapping. Analyses of multiple layers within tablets enabled a more reliable understanding of the variability in API domain sizes with respect to the independent variables. The relationship between API domain sizes and the manufactured tablets' content uniformity (CU) was also investigated using near-infrared transmission spectroscopy. Generally, at low shear, as the API particle size decreased, the frequency and size of API agglomerates increased, resulting in poor CU. However, in all cases, API domain sizes drastically reduced at high shear, resulting in an acceptable CU. The results of this work clearly demonstrated the utility of a multi-layer, multi-tablet, and high-resolution Raman chemical mapping as an off-line process analytical technology (PAT) system, to enable quality-by-design driven formulation and process development.

Online Detection and Classification of Moldy Core Apples by Vis-NIR Transmittance Spectroscopy

Apple moldy core disease is a common internal fungal disease. The online detection and classification of apple moldy core plays a vital role in apple postharvest processing. In this paper, an online non-destructive detection system for apple moldy core disease was developed using near-infrared transmittance spectroscopy in spectral range of 600–1100 nm. A total of 120 apple samples were selected and randomly divided into a training set and a test set based on the ratio of 2:1. First, basic parameters for detection of apples with moldy core were determined through detection experiments of samples in a stationary state. Due to the random distribution of the diseased tissue inside diseased apples, stationary detection cannot accurately identify the diseased tissue. To solve this problem, the spectra of apples in motion state transmitted forward by the transmission line were acquired. Three placement orientations of the apple in the carrying fruit cup were tested to explore the influence of fruit orientation on spectral characteristics and prediction. According to the performance of the model, the optimal preprocessing method and modeling method were determined (fixed orientation model and arbitrary orientation model). SPA was used to select the characteristic wavelengths to further improve the online detection speed. The overall results showed that the multi-spectra model using mean spectra of three orientations was the best. The prediction accuracies of multi-spectra model using SPA for three orientations were 96.7%, 97.5% and 97.5% respectively. As a conclusion, the arbitrary orientation model was beneficial to improve the online detection of apple moldy core disease.

Evaluation of optical properties of tofu samples produced with different coagulation temperatures and times using near-infrared transmission spectroscopy

To investigate the feasibility of using near-infrared (NIR) spectroscopy to evaluate the coagulation extent of tofu, the optical properties of tofu samples obtained with different coagulation temperatures and times were measured in the wavelength range 1300–1800 nm. The apparent absorbance was measured using the NIR transmission spectrum, and then divided into absorption and scattering components by curve fitting. The apparent absorbance increased over the entire wavelength range as the coagulation temperature increased, in which the scattering increase was dominant over that of the absorption. The viscosity was measured as an index of the physical properties of tofu, which increased with the coagulation temperature and had a high correlation (R2 = 0.803) with one of the fitting parameters of the scattering. The microstructure of the tofu became finer as the coagulation temperature increased, which suggests that the increase in scattering might be due to changes in the microstructure. Furthermore, the light scattering increased with the coagulation time as well as with the coagulation temperature. The increase in scattering with the coagulation time followed the first-order reaction kinetics. Both scattering intensity and increase speed were larger at shorter wavelengths. This can be explained by the theory of Mie scattering and multiple scattering, indicating that scattering measurements at shorter wavelengths are useful for monitoring of coagulation. Thus, light scattering measurements using NIR spectroscopy are promising for investigating the physical properties of tofu.

Rapid determination of ammonia nitrogen concentration in biogas slurry based on NIR transmission spectroscopy with characteristic wavelength selection

As an important intermediate product of the anaerobic digestion (AD) process, ammonia nitrogen (AN) is an essential indicator for analyzing the nitrogen nutrition level in the biogas system. To realize rapid evaluation of ammonia nitrogen concentration (ANC) in biogas slurry during AD, a detection model of ANC was established by combining near infrared (NIR) transmission spectroscopy with partial least squares. The CARS-GSA was constructed based on the competitive adaptive reweighted sampling (CARS) algorithm combined with the genetic simulated annealing algorithm (GSA), which was applied to characteristic wavelength selection of AN for improving the accuracy and efficiency of the spectral detection model. Through the selection of wavelengths by CARS-GSA, the modeling wavelengths decreased to 47 from 1557 of raw spectra, and the performance of the ANC regression model was effectively improved. For the CARS-GSA model, the determination coefficient, root mean squared error and residual predictive deviation of the validation set were 0.989, 12.549 and 9.662, respectively. For the independent test set, the above indexes were 0.990, 21.341 and 9.821, and consistent with those of the calibration set. The results showed that CARS-GSA could effectively reduce modeling wavelength variables and enhance prediction ability of the model. This study provides a theoretical support for the application of NIR transmission spectroscopy in online detection of ANC in biogas slurry.

Optimal wavelengths selection from all points for blood species identification based on spatially resolved near-infrared diffuse transmission spectroscopy

Blood species identification is of great significance for forensic science, wildlife protection. In our previous research, we proved spatially resolved near-infrared transmission spectroscopy method was feasible and promising for non-contact blood species identification, including chicken, goat, macaque, pig and rat, with samples measured in the cuvettes. However, spatially resolved spectra may bring crosstalk between the contributing chromophores while increasing effective information. In this paper, we explored wavelength selection methods for spatially resolved transmission spectroscopy from two different views—selecting variables from wavebands of a single point (or position) and selecting variables from all wavebands of all points (or positions). Equidistant Combination Partial Linear Square (ECPLS) method was used for wavelength selection. Results showed that the PLSDA model built with the wavelengths selected by ECPLS methods from different view of points both had a better performance than the PLSDA model built with full wavebands. We further test the reliability of the wavelengths selected by ECPLS methods from different view of points with an external validation. The results indicated that variables selected from all points was more reliable than from single point, and also full wavebands. Moreover, the result can also direct the collection of spatially resolved spectra with proper scanning position. The overall results sufficiently demonstrate the PLSDA model constructed using selected wavelength variables by a proper wavelength variable method can be more effective and accurate. And the results can guide the design of the measurement system for spatially resolved spectroscopy.

Time dependent berry maturation for planting density levels in Coffea arabica L. beans: Mixture design-fingerprinting using near-infrared transmittance spectroscopy

A three-step metabolomics strategy, employing near-infrared transmittance spectroscopy coupled with mixture design-fingerprints, chemometrics, and classical analytical assays, was used to determine metabolite abundances in coffee beans originating from three periods of berry maturation at the same of Coffea arabica plants cultivated at two planting densities. A first step screening showed that ethanol/dichloromethane and ethanol/dichloromethane/acetone extracts contained chemical information permitting classifications of beans cultivated at different planting densities whereas the acetone extract permitted the discrimination of time dependent berry maturation. In the second phase, selected extracts originating from different agronomical management conditions indicated that the 1680, 1660, and 1090 nm bands of the bean fingerprints were associated with these chemical changes. Finally, the third step consisted of classical analytical quantifications of caffeine, carbohydrates, lipids, and chlorogenic acids. This validated most of the conclusions obtained from the PCA score and loading exploratory analysis. The planting density and the harvest period systematic PCA-score patterns indicated gradual changes in the chemical profiles of coffee beans from the first to the third harvest. Partial least squares-discriminant analysis and ANOVA-simultaneous component analysis confirmed the classification and statistical significance of the agronomical management main effects and their interactions.

Non-Destructive Identification and Estimation of Granulation in Honey Pomelo Using Visible and Near-Infrared Transmittance Spectroscopy Combined with Machine Vision Technology

Granulation is a physiological disorder of juice sacs in citrus fruit, causing juice sacs to become hard and dry and resulting in decreased internal quality of citrus fruit. Honey pomelo is a thick-skinned citrus fruit, and it is difficult to identify the extent of granulation by observation of the outer peel and fruit shape. In this study, a rapid and non-destructive testing method using visible and near-infrared transmittance spectroscopy combined with machine vision technology was applied to identify and estimate granulation inside fruit. A total of 600 samples in different growth periods was harvested, and fruit were divided into five classes according to five granulation levels. Spectral data were obtained for two ranges of 400–1100 nm and 900–1700 nm by visible and near-infrared transmittance spectroscopy. In addition, chemometrics were used to measure the chemical changes of soluble solid content (SSC), titratable acidity (TA), and moisture content (MC) caused by different granulation levels. Machine vision technology can rapidly estimate the external characteristics of samples and measure the physical changes in mass and volume caused by different granulation levels. Compared with using a single or traditional methods, the predictive performances of multi-category classification models (PCA-SVM and PCA-GRNN) were significantly enhanced. In particular, the model accuracy rate (ARM) was 99% for PCA-GRNN, with classification accuracy (CA), classification sensitivity (CS), and classification specificity (CSP) of 0.9950, 0.9750, and 0.9934, respectively. The results showed that this method has great potential for the identification and estimation of granulation. Multi-source data fusion and application of a multi-category classification model with the smallest number of input layers and acceptable high predictive performances are proposed for on-line applications. This method can be effectively used on-line for the non-destructive detection of fruits with granulation.