Quantitative Evaluation of Color, Firmness, and Soluble Solid Content of Korla Fragrant Pears via IRIV and LS-SVM

Abstract

Customers pay significant attention to the organoleptic and physicochemical attributes of their food with the improvement of their living standards. In this work, near infrared hyperspectral technology was used to evaluate the one-color parameter, a*, firmness, and soluble solid content (SSC) of Korla fragrant pears. Moreover, iteratively retaining informative variables (IRIV) and least square support vector machine (LS-SVM) were applied together to construct evaluating models for their quality parameters. A set of 200 samples was chosen and its hyperspectral data were acquired by using a hyperspectral imaging system. Optimal spectral preprocessing methods were selected to obtain out partial least square regression models (PLSRs). The results show that the combination of multiplicative scatter correction (MSC) and Savitsky-Golay (S-G) is the most effective spectral preprocessing method to evaluate the quality parameters of the fruit. Different characteristic wavelengths were selected to evaluate the a* value, the firmness, and the SSC of the Korla fragrant pears, respectively, after the 6 iterations. These values were obtained via IRIV and the reverse elimination method. The correlation coefficients of the validation set of the a* value, the firmness, and the SSC measure 0.927, 0.948, and 0.953, respectively. Furthermore, the values of the regression error weight, γ, and the kernel function parameter, σ2, for the same parameters measure (8.67 × 104, 1.21 × 103), (1.45 × 104, 2.93 × 104), and (2.37 × 105, 3.80 × 103), respectively. This study demonstrates that the combination of LS-SVM and IRIV can be used to evaluate the a* value, the firmness, and the SSC of Korla fragrant pears to define their grade.

Study on Qualitative Impact Damage of Loquats Using Hyperspectral Technology Coupled with Texture Features

“…Based on the differences of kernel function, and band selection methods, the LS-SVM models were established based on spectra (S-LS-SVM), and spectra combined with texture features (S-T-LS-SVM). The kernel functions included the radial basis function (RBF), and linear kernel function (Lin) [ 37 ]. As shown in Table 1 , the overall accuracy of the LS-SVM model based on RBF (RBF-LS-SVM) was better than the LS-SVM model based on the LIN kernel function (LIN-LS-SVM).…”

Recent Innovations in Post-Harvest Preservation and Protection of Agricultural Products

“…Results demonstrated that the combination of IRIV and LS-SVM can be used to predict values for color parameter, a *, firmness, and SSC to define grade of Korla fragrant pears with correlation coefficients of the validation set measuring 0.927, 0.948, and 0.953, respectively [8].…”

Construction method of quantitative evaluation model for the maturity of Korla fragrant pear

“…Pericarp color is often viewed as an important index for the maturity of Korla fragrant pears, and it importantly influences the commodity value of fruits [28] . Figures 1f-1h show that the values of L * , a * , and b * achieved uniform growths with increased accumulated temperature.…”