Search

• Multisensor data-fusion approach was proposed for fusion of NIR and MIR datasets of tuber flours samples. • Data-fusion model named SOPLS showed significant improvement in the prediction accuracy as compared to using a single sensor model. • Proposed methodology to be applicable for improving consistency, robustness and precision while reducing uncertainty in the prediction model. This study evaluates the near-infrared spectroscopy (NIR) and mid-infrared spectroscopy (MIR) complementary spectral ranges to predict six different quality traits, which include chemical components such as amylose, starch, protein, glucose, cellulose, and moisture contents, in tubers and root flours. The sequential orthogonalized partial least square regression (SOPLS), a recently developed multi-sensor data-fusion approach, was adapted to improve the performance of the model in predicting the chemical properties of the flour samples. Furthermore, the performance of the SOPLS model was compared to that of traditional PLS modeling. Compared to the earlier results acquired using individual sensor modeling (with the traditional PLS model), the SOPLS fusion model showed significant improvement in the prediction performance for all cases except glucose. Particularly, the highest improvement in performance was observed for the prediction of cellulose, showing a 22.8% increase in coefficient of determination for prediction ( R 2 p) and 66.5% decrease in root mean square of prediction (RMSEP) values. Therefore, we concluded that the data-fusion approach used in this study exhibited better performance compared to the model using individual sensors. Furthermore, the multi-sensor fusion with the sequential approach is not limited to NIR and MIR data only and can be used for complementary information fusion to further improve the performance of the model. [ABSTRACT FROM AUTHOR]

A near-infrared (NIR) hyperspectral imaging (HSI) system was used to predict viability and vigor (in term of germination periods) in muskmelon seeds. Hyperspectral images of muskmelon seeds were acquired using a NIR push-broom HSI system covering the spectral range of 948–2494 nm. After NIR spectra collection, all seeds underwent a germination test to confirm their viability and vigor. The spectra from seeds with 3 and 5 germination days and nongerminated seeds were further used for development of a classification model of partial least-squares discriminant analysis (PLSDA). Most effective wavelengths were selected using three model-based variable selection methods, i.e., variable important in projection (VIP), selectivity ratio (SR), and significance multivariate correlation (sMC), which selected 23, 18, and 19 optimal variables, respectively, from full set of 208 variables. The selected variables from different waveband selection methods were found genuine and significant for interpreting the prediction results of seed viability and vigor. Subsequently, the PLS-DA model was constructed using individual VIP-, SR-, or sMC-selected variables. The results demonstrated that the PLSDA model developed with the selected optimal variables from the different methods provided comparable results for the calibration set; however, the PLSDA-SR method afforded the highest classification accuracy (94.6%) for a validation set used to determine the viability and vigor of muskmelon seeds. The wavelengths selected by the different methods represents chemical components of the seed and the attribute of germination ability was chosen most often. [ABSTRACT FROM AUTHOR]

Aflatoxins are toxic metabolites produced by the fungi Aspergillus flavus and Aspergillus parasiticus . They can contaminate a wide range of crops before harvest and during storage. Contaminated grains are associated with economic losses for cultivators as well as potential health hazards to both humans and animals. In this study, a short-wave infrared (SWIR) hyperspectral imaging technique was utilized to detect aflatoxin contamination on corn kernels. Corn samples were inoculated with four different aflatoxin B 1 (AFB1) concentrations (10, 100, 500 and 1000 μg/kg) while control samples were surface-disinfected with a PBS solution. Both infected and control samples were scanned with an SWIR hyperspectral system over the spectral range of 1100–1700 nm. A partial least squares discriminant analysis (PLS-DA) model was developed to categorize control and infected kernels and the highest overall classification accuracy yielded from the developed model was 96.9%. Spectral deviation was observed between the control and inoculated samples as the AFB1 concentrations increased. In addition, the contamination map generated with the PLS-DA model provided the visual appearance of infected samples. Our results suggest that SWIR hyperspectral imaging is a rapid, accurate, and non-destructive technique for the detection of toxic metabolites in grains and could be an alternative to manual techniques. [ABSTRACT FROM AUTHOR]