NON-DESTRUCTIVE PREDICTION OF SOLUBLE SOLID CONTENT IN KIWIFRUIT BASED ON VIS/NIR HYPERSPECTRAL IMAGING

Soluble solid content (SSC) is a major quality index of kiwifruits. Visible near-infrared (Vis/NIR) hyperspectral imaging with the genetic algorithm (GA) was adopted in this study to realize non-destructive prediction kiwifruit SSC. A laboratory Vis/NIR system established collect 120 samples at range 400–1100 nm. The average reflectance spectral data region interest were obtained after different preprocessing method, namely, Savitzky–Golay smoothing (SG), multiplicative scatter correction (MSC), and their combination method. models partial least squares regression, multiple linear support vector machine (LS-SVM) built for determining SSC by using effective feature wavelength variables selected GA, respectively. results show that SG+MSC best precisions GA are higher than full data. GA-LS-SVM model has performance correlation coefficient (R=0.932) standard error (SEP=0.536° Bx) predicting accuracy been improved 5.6% compared full-band This provides an method detection