The relationship between optical properties and soluble solid contents of Gong pear for non-destructive internal quality inspection

Abstract

To investigate the relationship between optical properties and soluble solids content (SSC) in different regions of Gong pear. The optical properties of the upper, middle and lower parts of Gong pear were measured by the single integrating sphere system in the range of 500–1100 nm. The differences of absorption coefficient (${\mu }_{a}$) spectra and reduced scattering coefficient (${\mu }_{s}^{\prime}$) spectra in three regions of Gong pear were analyzed. The differences of ${\mu }_{a}$ and ${\mu }_{s}^{\prime}$ at 670,710, 750, 800 and 980 nm were analyzed. Gong Pear SSC was positively correlated with ${\mu }_{a}$ and ${\mu }_{s}^{\prime}$ in the range of 500–1050 nm.The local model of Gong pear SSC was established based on the ${\mu }_{a}$ spectra and ${\mu }_{s}^{\prime}$ spectra in the upper, middle and lower regions respectively. And the partial least square regression (PLSR) model and support vector regression (SVR) model were established based on the mean ${\mu }_{a}$ spectra, mean ${\mu }_{s}^{\prime}$ spectra and mean ${\mu }_{a}$+${\mu }_{s}^{\prime}$ spectra after standard normal variables (SNV) pretreatment. The results showed that the optical properties of the upper, middle and lower sections of Gong pear were less different. Among all established SSC prediction models, the model based on the mean ${\mu }_{a}$ spectra had the best prediction effect. Its correction coefficient of determination (${R}_{c}^{2}$) and prediction coefficient of determination (${R}_{P}^{2}$) were 0.894 and 0.837, and its correction root mean square error (RMSEC) and prediction root mean square error (RMSEP) were 0.305 and 0.429, respectively. The results showed that soluble solids content mainly affected the absorption characteristics of Gong pear, and the internal quality of Gong pear could be predicted better based on the absorption coefficient.