Search

Nutrients derived from fish feed are insufficient for optimal plant growth in aquaponics; therefore, they need to be supplemented. Thus, estimating the amount of supplementation needed can be achieved by looking at the nutrient contents of the plant. This study aims to develop trustworthy machine learning models to estimate the nitrogen (N), phosphorus (P), and potassium (K) contents of aquaponically grown lettuce. A FieldSpec4, Pro FR portable spectroradiometer (ASD Inc., Analytical Spectral Devices Boulder, Boulder, CO, USA) was used to measure leaf reflectance spectra, and 128 lettuce seedlings given four NPK treatments were used for spectra acquisition and total NPK estimation. Principal component analysis (PCA), genetic algorithms (GA), and sequential forward selection (SFS) were applied to select the optimal wavebands. Partial least squares regression (PLSR), back-propagation neural network (BPNN), and random forest (RF) approaches were used to develop the predictive models of NPK contents using the selected optimal wavelengths. Good and significantly correlated predictive accuracy was obtained in comparison with the laboratory-measured freshly cut lettuce leaves with R2 ≥ 0.94. The proposed approach provides a pathway toward automatic nutrient estimation of aquaponically grown lettuce. Consequently, aquaponics will become more intelligent, and will be adopted as a precision agriculture technology. [ABSTRACT FROM AUTHOR]

The total volatile basic nitrogen (TVB-N) content is representative index for measuring the freshness of cooked meat. The study investigated integrating spectral and image information from visible and near-infrared hyperspectral imaging for predicting TVB-N values in cooked beef during storage. Nine optimal wavelengths were selected by uninformative variable elimination (UVE) and successive projections algorithm (SPA). Thirty-six singular values as texture features were extracted using discrete wavelet transform (DCT). Partial least squares regression (PLSR) and least squares-support vector machine (LS-SVM) models were established using spectral, image, and their fusion information. The models based on data fusion yielded satisfactory prediction results: PLSR gave R c 2 and R P 2 values of 0.9512 and 0.9401 with RMSEC and RMSEP of 1.9037 and 1.8942; LS-SVM gave R c 2 and R P 2 values of 0.9674 and 0.9579 with RMSEC and RMSEP of 1.6538 and 1.5435. The distribution of TVB-N values from sample images was visualized using the fused LS-SVM model with image processing algorithms. This indicated that integrating spectral and image information from hyperspectral imaging coupled with the LS-SVM or PLSR algorithm could predict the TVB-N value and visualize its distribution in cooked beef. [ABSTRACT FROM AUTHOR]

Hyperspectral imaging (HSI) is a promising technique to evaluate food quality in a rapid and non-destructive way. Several types of HSI sensors are now commercially available. The aim of the current study was to compare one single shot and two line-scanning based HSI sensors in different spectral regions by using two wavelength dispersive devices – a prism-grating-prism monochromator and a mosaic wavelength dispersion unit. The research sample was cured pork meat, which is a popular meat product due to its flavor and long shelf life. Three main chemical compositions of cured pork meat – moisture, protein and fat content – were evaluated. The spectral properties of the three sensors were employed to establish partial least squares regression (PLSR) predictive models by applying different optimal wavelengths selection strategies. The results showed that optimal wavelength selection was more important than the selection of the wavelength dispersion device. • Three hyperspectral imaging (HSI) systems were compared. • Prediction models for compositions of dry and wet cured pork were established. • Nine wavelengths were sufficient for predicting chemical compositions. • Slight differences between the two wavelengths dispersion systems were observed. • Optimal wavelength selection was a key step in improving models of HSI. [ABSTRACT FROM AUTHOR]

Wavelength selection from a large diffuse reflectance spectroscopy (DRS) dataset enables removal of spectral multicollinearity and thus leads to improved understanding of the feature domain. Feature selection (FS) frameworks are essential to discover the optimal wavelengths for tissue differentiation in DRS-based measurements, which can facilitate the development of compact multispectral optical systems with suitable illumination wavelengths for clinical translation. The aim was to develop an FS methodology to determine wavelengths with optimal discriminative power for orthopedic applications, while providing the frameworks for adaptation to other clinical scenarios. An ensemble framework for FS was developed, validated, and compared with frameworks incorporating conventional algorithms, including principal component analysis (PCA), linear discriminant analysis (LDA), and backward interval partial least squares (biPLS). Via the one-versus-rest binary classification approach, a feature subset of 10 wavelengths was selected from each framework yielding comparable balanced accuracy scores (PCA: 94.8 ± 3.47 % , LDA: 98.2 ± 2.02 % , biPLS: 95.8 ± 3.04 % , and ensemble: 95.8 ± 3.16 %) to those of using all features (100%) for cortical bone versus the rest class labels. One hundred percent balanced accuracy scores were generated for bone cement versus the rest. Different feature subsets achieving similar outcomes could be identified due to spectral multicollinearity. Wavelength selection frameworks provide a means to explore domain knowledge and discover important contributors to classification in spectroscopy. The ensemble framework generated a model with improved interpretability and preserved physical interpretation, which serves as the basis to determine illumination wavelengths in optical instrumentation design. [ABSTRACT FROM AUTHOR]

The objective of this study was to develop a prototype multispectral imaging system for online quality assessment on pomegranate fruit. At first, a visible/near infrared spectroscopy (400–1100 nm) was tested for non-destructive determination of total soluble solids, titratable acidity, and pH. The spectral data were analyzed using the partial least square analysis. Then to establish consistent multispectral imaging system, the highest absolute values of β-coeffcients correspond to wavelengths from the best partial least square calibration model were selected and used for identifying the optimal wavelengths. Consequently, a multispectral imaging system was developed based on the effective wavelengths 700, 800, 900, and 1000 nm. The performance of the developed multispectral imaging system was evaluated by multiple linear regression models. The multiple linear regression model predict total soluble solids withr= 0.97, root mean square error of calibration = 0.21°Brix, and ratio performance deviation = 6.7 °Brix. Also, the results showed that the models had good predictive ability for pH and titratable acidity. Results showed that the developed multispectral imaging system based on the optimal wavelengths could be used for online quality assessment of pomegranate fruit. [ABSTRACT FROM AUTHOR]

Muscle foods play an important role in providing a vital source of high-quality protein, amino acids and vitamin for human health. Chemical composition is one of the most vital information of muscle foods, which directly relates to the quality of pork, beef, chicken, fish and other meats. Therefore, it is significant to identify the chemical information of muscle foods for the purpose of controlling the quality and safety of meat. Hyperspectral imaging can obtain spectral and spatial information of targets simultaneously and has been developed for rapid and nondestructive determination and identification of chemical information of muscle foods. This review focuses on recent applications of hyperspectral imaging technology for the measurement and analysis of chemical composition of muscle foods, including moisture content, fat and fatty acid, pH, protein content, pigment, salt content and freshness attributes. The fundamentals of hyperspectral imaging as well as future development trends are also presented and discussed. [ABSTRACT FROM AUTHOR]

In this paper, it is shown how to select the optimal wavelengths minimizing the relative error and the standard deviation of the temperature. Furthermore, it is shown that the optimal wavelengths in mono-spectral and bi-spectral methods (for a Planck’s law) can be determined by laws analogous to the displacement Wien’s law. The simplicity of these laws can thus allow real-time selection of optimal wavelengths for a control/optimization of industrial processes, for example. A more general methodology to obtain the optimal wavelengths selection in a multi-spectral method (taking into account the spectral variations of the global transfer function including the emissivity variations) for temperature measurement of surfaces exhibiting non-uniform emissivity, is also presented. This latter can then find an interest in glass furnaces temperature measurement with spatiotemporal non-uniformities of emissivity, the control of biomass pyrolysis, the surface temperature measurement of buildings or heating devices, for example. The goal consists of minimizing the standard deviation of the estimated temperature (optimal design experiment). For the multi-spectral method, two cases will be treated: optimal global and optimal constrained wavelengths selection (to the spectral range of the detector, for example). The estimated temperature results obtained by different models and for different number of parameters and wavelengths are compared. These different points are treated from theoretical, numerical and experimental points of view. [ABSTRACT FROM AUTHOR]

A hyperspectral imaging system was developed in the spectral region between 400 and 1,100 nm to investigate its potential for predicting the shelf life of bananas with different browning levels. Principal component analysis (PCA) was conducted for reducing data dimensionality and selecting optimal wavelengths. Five optimal wavelengths (454, 486, 559, 686, and 728 nm) were found. Among all principal component (PC) images, PC-4 was selected to segregate browning area from normal surface using a simple threshold algorithm for extracting image features of browning area. The average spectra were obtained by calculating the mean of the spectra of all browning areas in the hyperspectral images of bananas. Then, image features and average spectra in five optimal wavelengths were used to develop classification models for predicting the shelf life of banana samples by determining their browning levels using back propagation (BP), radial basis function (RBF), and self-organizing feature maps (SOM) networks. Results indicated that the classification models using both image features and average spectra were obviously superior to those models using image features or average spectra alone. Among all classifiers, BP classifier had the best performance with the best classification rates of 95.6 % for training set and 90.5 % for testing set, respectively. The results demonstrated that hyperspectral imaging has great potential in predicting banana shelf life based on combination of image features and average spectra. [ABSTRACT FROM AUTHOR]